JP2004272557A - Area separation processing method, area separation processing program, recording medium storing the same, image processing apparatus, and image forming apparatus having the same - Google Patents

Abstract

An object of the present invention is to provide an area separation processing method, an area separation processing program, and a recording medium on which the area separation processing method and the recording medium recording the same can be performed with high accuracy and high speed. Provided is a processing device and an image forming apparatus including the same.
A background pixel is detected by a background detection processing unit of a block determination unit (b1). The horizontal edge component is calculated by the horizontal edge component calculation processing unit 111, and the vertical edge component is calculated by the vertical edge component calculation processing unit 112 (b2, 3). The edge direction classification processing unit 12 classifies each pixel into a non-edge pixel, a horizontal edge pixel, a vertical edge pixel, and a diagonal edge pixel (b4). The feature amount based on the correlation in the edge direction is calculated by the direction correlation calculation processing unit 13 (b5). Then, the determination processing unit 14 determines whether the determination block or the pixel of interest in the determination block belongs to any of the base, photograph, character, halftone, and indefinite regions (b6).
[Selection diagram] FIG.

Description

【００８６】
ＬＰＦの場合は、式（４）に示すように処理の対象となる画素群の画素データＯｒｉｇｉｎａｌＬとフィルタｆｉｌｔｅｒ＿ｈｏｒとの畳み込み演算を行い、これを元の画素群の画素データＯｒｉｇｉｎａｌＬから減算する。こうすることで、ノイズの影響を抑制しつつ、注目画素の水平エッジ成分のみを抽出することができる。
【００８７】
【数２】

【００８８】
なお、フィルタの形状はハイパス、ローパスのいずれかの性質を持つものであればどのようなものであってもよく、たとえばラプラシアンフィルタなどを用いることができる。
【００８９】
また、判定ブロックの端に位置する画素についてのフィルタ処理は、インターリーブ処理によって、たとえば判定ブロック内で隣接する画素の画素データを、判定ブロック外の画素データとして補間することで行っている。
【００９０】
次に、垂直エッジ成分算出処理（図５に示すステップｂ３）について説明する。図８は、垂直エッジ成分算出処理部１１２による垂直エッジ成分算出処理の手順を示すフローチャートである。ここでは、判定ブロックを構成する画素ごとに、その画素データの垂直エッジ成分を数値として算出し、算出した値を後述するエッジ方向分類処理で使用する。基本的な処理は水平エッジ成分処理と同様である。
【００９１】
ステップｅ１で、以下の式（５）に基づいて注目画素の垂直エッジ成分ｖｅｒを算出する。
ｖｅｒ＝Ｅｄｇｅ（ｆｉｌｔｅｒ＿ｖｅｒ，ＯｒｉｇｉｎａｌＬ）…（５）
【００９２】
ｆｉｌｔｅｒ＿ｖｅｒとは、垂直エッジ成分抽出のためのフィルタであり、ｆｉｌｔｅｒ＿ｖｅｒとｆｉｌｔｅｒ＿ｈｏｒは以下の式（６）の関係を満たすことが好ましい。
ｆｉｌｔｅｒ＿ｖｅｒ＝ｆｉｌｔｅｒ＿ｈｏｒ^Τ …（６）
【００９３】
ここでの「Ｔ」は、転地演算を表す。そして、ステップｅ２ではステップｂ３へ戻る。
【００９４】
次に、エッジ方向分類処理（図５に示すステップｂ４）について説明する。図９は、エッジ方向分類処理部１２によるエッジ方向分類処理の詳細な手順を示すフローチャートである。ここでは、先に算出した水平エッジ成分ｈｏｒおよび垂直エッジ成分ｖｅｒに基づいて、判定ブロックの各画素を非エッジ画素、水平エッジ画素、垂直エッジ画素、あるいは斜めエッジ画素のいずれかに分類するとともに、判定ブロック内のエッジ画素の総数を算出する。この処理は、判定ブロック内上段の左端の画素から右端の画素の順に行い、これを各段で繰り返すことで、判定ブロック内の左上の画素から右下の画素にかけて行う。
【００９５】
まず、ステップｆ１では、注目画素はエッジ画素か否かが判断される。この判断は、水平エッジ成分ｈｏｒおよび垂直エッジ成分ｖｅｒの二乗和（ｈｏｒ^２＋ｖｅｒ^２）を算出し、これを閾値Ｔ２と比較することによって行う。なお、閾値Ｔ２との比較にそれぞれのエッジ成分の二乗和（ｈｏｒ^２＋ｖｅｒ^２）を用いているのは、水平エッジ成分、垂直エッジ成分の算出結果が負の値になる場合があるからであり、単なるエッジ成分の和（ｈｏｒ＋ｖｅｒ）では、注目画素周辺の濃度変化を正しく評価できないためである。
【００９６】
（ｈｏｒ^２＋ｖｅｒ^２）が閾値Ｔ２より大きい場合には、注目画素周辺の濃度変化が大きく、注目画素をエッジ画素と判定してステップｆ３に進む。（ｈｏｒ^２＋ｖｅｒ^２）が閾値Ｔ２以下の場合には、ステップｆ２に進む。ステップｆ２では、注目画素周辺の濃度変化が小さいので、非エッジ画素と判定され、ステップｆ９へ進む。
【００９７】
ステップｆ３では、エッジ総数Ｃｎｔ＿ｔｏｔａｌに１が加算される。エッジ総数Ｃｎｔ＿ｔｏｔａｌとは、判定ブロック内におけるエッジ画素の総数である。ステップｆ４では、水平エッジ成分ｈｏｒと垂直エッジ成分ｖｅｒの二乗の差分値（ｈｏｒ^２−ｖｅｒ^２）が閾値Ｔ３と比較される。（ｈｏｒ^２−ｖｅｒ^２）が閾値Ｔ３より大きい場合には、ステップｆ５へ進み、注目画素は水平エッジ画素と判定され、ステップｆ９に進む。（ｈｏｒ^２−ｖｅｒ^２）が閾値Ｔ３以下の場合には、ステップｆ６に進む。
【００９８】
ステップｆ６では、（ｈｏｒ^２−ｖｅｒ^２）と、閾値Ｔ３より小さい値の閾値Ｔ４との比較が行われ、（ｈｏｒ^２−ｖｅｒ^２）が閾値Ｔ４より小さい場合には、ステップｆ７に進む。ステップｆ７では、注目画素は垂直エッジ画素と判定され、ステップｆ９に進む。（ｈｏｒ^２−ｖｅｒ^２）が閾値Ｔ４以上の場合には、ステップｆ８に進む。ステップｆ８では、注目画素は斜めエッジ画素と判定され、ステップｆ９に進む。このように、閾値Ｔ３との比較にそれぞれのエッジ成分の二乗差を用いているのは、水平エッジ成分、垂直エッジ成分の算出結果が負の値になる場合があり、単なるエッジ成分の単なる差（ｈｏｒ−ｖｅｒ）では正しいエッジの方向を評価できないからである。
【００９９】
ステップｆ９では、判定ブロックを構成する全ての画素について、エッジ方向分類処理が終了したか否かが判定される。終了していなければステップｆ１に戻り、終了していればエッジ方向分類処理を終了する。
【０１００】
このように、エッジ方向の決定には、水平エッジ成分ｈｏｒの２乗と垂直エッジ成分ｖｅｒの２乗との差分を算出するだけでよいので、たとえばエッジ角度の算出に水平成分と垂直成分の比の逆正接を用いる場合と比較して、格段に計算量を削減することができ、精度を下げることなく処理の高速化を図ることができる。また、当該処理を行う回路の規模増大を抑えることができる。
【０１０１】
このエッジ方向分類処理は次の手順で行ってもよい。
図１０は、角度によるエッジ方向の分類方法を表す図である。画像データの水平ライン方向を基準とした場合、水平方向とエッジの成す角度をθ（０°≦θ≦９０°）とし、θがφ_Ｈよりも小さければ水平エッジ（θ＜φ_Ｈ）、φ_Ｖよりも大きければ垂直エッジ（θ＞φ_Ｖ）、それ以外（φ_Ｈ≦θ≦φ_Ｖ）であれば斜めエッジと定義すると、以下の式（７）〜（９）のような判定式によって、エッジの方向を判定することができる。具体的には、エッジ角度が１５度未満は水平エッジ（θ＜φ_Ｈ＝１５°）、７５度を超えるときは垂直エッジ（θ＞φ_Ｖ＝７５°）というように、予め定めておけばエッジ方向を精度良く分類することができる。
ｖｅｒ^２−ｈｏｒ^２ｔａｎ^２φ_Ｈ＜０（Ｈｏｒｉｚｏｎｔａｌ；水平）…（７）
ｖｅｒ^２−ｈｏｒ^２ｔａｎ^２φ_Ｖ＞０（Ｖｅｒｔｉｃａｌ；垂直） …（８）
ｏｔｈｅｒｗｉｓｅ （Ｄｉａｇｏｎａｌ；斜め） …（９）
【０１０２】
図１１は、エッジ方向分類処理の他の手順を示すフローチャートである。処理手順は、図９に示す処理手順と略同様であるが、ステップｆ４’およびステップｆ６’における、水平エッジ、垂直エッジおよび斜めエッジを判定するための判断基準が異なる。すなわち、判断基準に、前記の式（７）、（８）を用いるのである。
【０１０３】
次に、方向相関算出処理（図５に示すステップｂ５）について説明する。図１２は、方向相関算出処理部１３による方向相関算出処理の詳細な手順を示すフローチャートである。相関性とは、各画素のエッジの方向と近傍画素のエッジの方向とがどれくらい同じかまたは異なっているか、すなわち、近傍画素のうち同じエッジの方向を持つ近傍画素の割合を示す。本実施形態では、３×３画素の画素群の中心を注目画素とし、その周囲の８画素を近傍画素とする。
【０１０４】
判定ブロック内の各エッジ画素を注目画素としたときに、相関性として注目画素の近傍エッジ画素のうち、注目画素のエッジ方向と同じエッジ方向を有する近傍エッジ画素数を注目画素ごとに計数し、計数された近傍エッジ画素数が所定の値より少ない注目画素の画素数を特徴量として算出する。この処理は、判定ブロック内上段の左端の画素から右端の画素の順に行い、これを各段で繰り返すことで、判定ブロック内の左上の画素から右下の画素にかけて行う。なお、判定ブロックの端に位置する画素についての処理は、インターリーブ処理によって、たとえば判定ブロック内で隣接する画素の画素データを補間することで行っている。
【０１０５】
まず、ステップｇ１では、エッジ方向分類処理のステップｆ１，ｆ２の判断結果に基づいて、注目画素がエッジ画素であるか否かが判断される。エッジ画素であればステップｇ２に進み、エッジ画素でなければステップｇ９に進む。
【０１０６】
ステップｇ２では、ステップｇ１と同様に、注目画素の近傍画素のうち所定の近傍画素がエッジ画素であるか否かが判定される。エッジ画素であればステップｇ３に進み、エッジ画素でなければステップｇ６に進む。ステップｇ３では、近傍エッジ数Ｃｎｔ＿ｎｅｉｇｈｂｏｒに１が加算される。これによって、エッジ画素である近傍画素の数を算出することができる。
【０１０７】
ステップｇ４では、エッジ方向分類処理におけるステップｆ４〜ｆ８の判断結果に基づいてエッジ方向を比較し、注目画素と近傍画素とのエッジの方向が等しいか否かが判定される。等しければステップｇ５に進み、等しくなければステップｇ６に進む。ステップｇ５では、同方向エッジ数Ｃｎｔ＿ｓａｍｅに１が加算される。これによって、注目画素のエッジと同方向のエッジを有する近傍画素の数を算出することができる。
【０１０８】
そして、ステップｇ６では、すべての近傍画素について処理が終了したか否かが判定される。終了していればステップｇ７に進み、終了していなければステップｇ２に戻り、次の近傍画素の処理を行う。
【０１０９】
ステップｇ７では、同方向エッジ数Ｃｎｔ＿ｓａｍｅを２倍したものと、注目画素の近傍エッジ数Ｃｎｔ＿ｎｅｉｇｈｂｏｒとが比較される。２Ｃｎｔ＿ｓａｍｅ＜Ｃｎｔ＿ｎｅｉｇｈｂｏｒとなった場合、同方向エッジ数が近傍エッジ数の半数より下回ることになるので、注目画素のエッジ方向と異なるエッジ方向を有する近傍エッジ画素の割合が多いことになる。ステップｇ８では、特徴量である相関画素数Ｃｎｔ＿ｈｉｇｈ＿ｃｏｒに１が加算される。そして、Ｃｎｔ＿ｓａｍｅおよびＣｎｔ＿ｎｅｉｇｈｂｏｒの値はクリアされる。ステップｇ９では、判定ブロックを構成する全ての画素について処理が終了したか否かが判断される。終了していれば方向相関算出処理を終了し、終了していなければステップｇ１に戻る。
【０１１０】
このように、画素ブロック内の注目画素ごとに、注目画素のエッジの方向と同じエッジの方向を有する近傍エッジ画素数を計数し、この計数値が所定の値より小さくなる注目画素の画素数を算出するので、画素ブロックの特徴量を容易に算出することができる。
【０１１１】
なお、方向相関算出処理は、以下のように内積を利用して行うこともできる。図１３は、内積を利用して方向相関算出処理を行う場合の処理手順を示すフローチャートである。まず、予めそれぞれのエッジの方向に対して以下のベクトルを割り当てておく。注目画素をＸｃ、近傍画素をＸｎとし、Ｘｃ，Ｘｎにはエッジの方向に対して割り当てたベクトル値（１０）〜（１３）を格納する。ここでのベクトル値が、エッジの方向に応じて置き換えられた所定の値となる。
水平方向→［１ ０ ０］ …（１０）
垂直方向→［０ １ ０］ …（１１）
斜め方向→［０ ０ １］ …（１２）
非エッジ→［０ ０ ０］ …（１３）
【０１１２】
これらのベクトル値を用いて、式（１４）および式（１５）から近傍エッジ数Ｃｎｔ＿ｎｅｉｇｈｂｏｒおよび同方向エッジ数Ｃｎｔ＿ｓａｍｅを算出する。
【０１１３】
【数３】

【０１１４】
【数４】

【０１１５】
具体的なフローでは、まずステップｈ１で、近傍エッジ数Ｃｎｔ＿ｎｅｉｇｈｂｏｒ＝Ｃｎｔ＿ｎｅｉｇｈｂｏｒ＋Ｘｎ・Ｘｎを算出し、ステップｈ２は、同方向エッジ数Ｃｎｔ＿ｓａｍｅ＝Ｃｎｔ＿ｓａｍｅ＋Ｘｃ・Ｘｎを算出し、これを画素ブロック内の全画素について繰り返す。
【０１１６】
ステップｈ３では、すべての近傍画素について処理が終了したか否かが判定される。終了していればステップｈ４に進み、終了していなければステップｈ１に戻る。ステップｈ４では、同方向エッジ数Ｃｎｔ＿ｓａｍｅを２倍したものと、注目画素の近傍エッジ数Ｃｎｔ＿ｎｅｉｇｈｂｏｒとが比較される。２Ｃｎｔ＿ｓａｍｅ＜Ｃｎｔ＿ｎｅｉｇｈｂｏｒであればステップｈ５に進み、２Ｃｎｔ＿ｓａｍｅ≧Ｃｎｔ＿ｎｅｉｇｈｂｏｒであればステップｈ６に進む。
【０１１７】
ステップｈ５では、相関画素数Ｃｎｔ＿ｈｉｇｈ＿ｃｏｒに１が加算される。そして、Ｃｎｔ＿ｓａｍｅおよびＣｎｔ＿ｎｅｉｇｈｂｏｒの値はクリアされる。ステップｈ６では、判定ブロックを構成する全ての画素について処理が終了したか否かが判断される。終了していれば方向相関算出処理を終了し、終了していなければステップｈ１に戻る。
【０１１８】
このように、方向相関算出処理に内積を利用することで、処理手順中の判断処理を減らし、処理を単純化することができる。すなわち、図１２に示したステップｇ１，ｇ２，ｇ４において分岐処理を行う必要がなくなるため、方向相関算出処理を高速に行うことができる。さらに、当該処理を行う回路の規模増大を抑えることができる。
【０１１９】
また、式（１０）〜（１３）のベクトル成分を各ビットに割り当て、水平エッジを「１００」、垂直エッジを「０１０」、斜めエッジを「００１」のように２進数で表現すれば、式（１４），（１５）の内積演算はビット演算により実行でき、さらに処理を高速に行うことができる。
【０１２０】
次に、判定処理（図５に示すステップｂ６）について説明する。判定処理は、判定ブロックごとに行い、対象となる判定ブロックは、下地画素検出処理、水平エッジ成分算出処理、垂直エッジ成分算出処理、エッジ方向分類処理および方向相関算出処理が終了した判定ブロックである。下地画素検出処理で算出した下地画素数Ｃｎｔ＿ｗｈｉｔｅ、エッジ方向分類処理で算出したエッジ画素の総数Ｃｎｔ＿ｔｏｔａｌ、方向相関算出処理で算出した高相関画素数Ｃｎｔ＿ｈｉｇｈ＿ｃｏｒを用いることで、当該判定ブロックが、下地ブロック、写真ブロック、文字ブロック、網点ブロックおよび不定ブロックのうちのいずれであるのかを判定する。
【０１２１】
図１４は、判定処理の詳細な手順を示すフローチャートである。まず、ステップｉ１では、下地画素数Ｃｎｔ＿ｗｈｉｔｅと閾値Ｔ５とが比較される。下地画素数Ｃｎｔ＿ｗｈｉｔｅが閾値Ｔ５よりも大きければステップｉ２に進み、下地画素数Ｃｎｔ＿ｗｈｉｔｅが閾値Ｔ５以下であればステップｉ３に進む。ステップｉ２では、当該判定ブロックを下地ブロックと判定し、判定処理を終了する。これは、判定ブロック内の下地画素の数が一定数より多く存在するからである。
【０１２２】
ステップｉ３では、エッジ画素の総数Ｃｎｔ＿ｔｏｔａｌと閾値Ｔ６とが比較される。エッジ画素の総数Ｃｎｔ＿ｔｏｔａｌが閾値Ｔ６よりも小さければ（Ｃｎｔ＿ｔｏｔａｌ＜閾値Ｔ６）ステップｉ４に進み、エッジ画素の総数Ｃｎｔ＿ｔｏｔａｌが閾値Ｔ６以上であれば（Ｃｎｔ＿ｔｏｔａｌ≧閾値Ｔ６）ステップｉ５に進む。ステップｉ４では、当該判定ブロックを写真ブロックと判定し、判定処理を終了する。これは、判定ブロック内のエッジ画素の総数が一定数より少なく、中間調画像であり、かつ、下地ブロックではないからである。
【０１２３】
ステップｉ５では、相関画素数Ｃｎｔ＿ｈｉｇｈ＿ｃｏｒと閾値Ｔ７とが比較される。Ｃｎｔ＿ｈｉｇｈ＿ｃｏｒが閾値Ｔ７よりも小さければ（Ｃｎｔ＿ｈｉｇｈ＿ｃｏｒ＜閾値Ｔ７）ステップｉ７に進み、Ｃｎｔ＿ｈｉｇｈ＿ｃｏｒが閾値Ｔ７以上であれば（Ｃｎｔ＿ｈｉｇｈ＿ｃｏｒ≧閾値Ｔ７）ステップｉ６に進む。
【０１２４】
ステップｉ７では、相関画素数Ｃｎｔ＿ｈｉｇｈ＿ｃｏｒを２倍した２Ｃｎｔ＿ｈｉｇｈ＿ｃｏｒとエッジ画素の総数Ｃｎｔ＿ｔｏｔａｌとが比較される。２Ｃｎｔ＿ｈｉｇｈ＿ｃｏｒがエッジ画素の総数Ｃｎｔ＿ｔｏｔａｌよりも小さければ（２Ｃｎｔ＿ｈｉｇｈ＿ｃｏｒ＜Ｃｎｔ＿ｔｏｔａｌ）、ステップｉ８に進む。この場合、相関画素数はエッジ画素の総数の半分よりも小さいことになり、判定ブロック内にはエッジの方向が同方向の画素が多く存在することになるので、ステップｉ８では、当該判定ブロックを文字ブロックと判定し、判定処理を終了する。
【０１２５】
ステップｉ７で、２Ｃｎｔ＿ｈｉｇｈ＿ｃｏｒがエッジ画素の総数Ｃｎｔ＿ｔｏｔａｌ以上であれば（２Ｃｎｔ＿ｈｉｇｈ＿ｃｏｒ≧Ｃｎｔ＿ｔｏｔａｌ）、ステップｉ９に進む。この場合、前記とは逆に判定ブロック内にはエッジの方向が異なる画素が多く存在することになるので、網点ブロックである可能性も考えられる。しかしながら、誤判定である可能性もあるのでこれを避ける理由から、ステップｉ９に進み、不定ブロックと判定し、判定処理を終了する。
【０１２６】
ステップｉ６でも同様に、相関画素数Ｃｎｔ＿ｈｉｇｈ＿ｃｏｒを２倍した２Ｃｎｔ＿ｈｉｇｈ＿ｃｏｒとエッジ画素の総数Ｃｎｔ＿ｔｏｔａｌとが比較される。２Ｃｎｔ＿ｈｉｇｈ＿ｃｏｒがエッジ画素の総数Ｃｎｔ＿ｔｏｔａｌよりも小さければ（２Ｃｎｔ＿ｈｉｇｈ＿ｃｏｒ＜Ｃｎｔ＿ｔｏｔａｌ）、判定ブロック内にはエッジの方向が同方向の画素が多く存在することになるので、文字ブロックである可能性も考えられる。しかしながら、誤判定である可能性もあるのでこれを避ける理由から、ステップｉ９に進み、不定ブロックと判定し、判定処理を終了する。
【０１２７】
２Ｃｎｔ＿ｈｉｇｈ＿ｃｏｒがエッジ画素の総数Ｃｎｔ＿ｔｏｔａｌ以上であれば（２Ｃｎｔ＿ｈｉｇｈ＿ｃｏｒ≧Ｃｎｔ＿ｔｏｔａｌ）、ステップｉ１０に進む。この場合、判定ブロック内には方向の異なるエッジが多く存在することになるので、ステップｉ１０では判定ブロックを網点ブロックと判定し、判定処理を終了する。
【０１２８】
以上の判定処理によって、判定ブロックを、文字ブロック、網点ブロック、写真ブロック、下地ブロック、不定ブロックに分離することができる。
【０１２９】
前述の各処理を画像データ中の全判定ブロックについて行うことで、画像データの領域分離処理を行うことができる。
【０１３０】
このように、画素のエッジの方向に基づく特徴量を用いることで、画像のエッジ方向という構造的な特徴に基づいた領域分離を行うことができる。これによって、周波数分布が非常に似ている領域であっても、精度よく分離することができる。
【０１３１】
以上では、所定の画素数からなる画素ブロックに対して領域分離を行う方法を示したが、画素ブロック内の注目画素に対して領域判定を行うようにしてもよい。
【０１３２】
図１５は、画像形成装置Ｙの構成を示すブロック図である。画像形成装置Ｙは、画像処理装置Ｘ、画像入力装置６０および画像出力装置６１を備え、たとえば、電子写真方式やインクジェット方式を用いたデジタル複写機もしくは複合機などとして利用される。
【０１３３】
前述の領域分離処理部Ａは、画像処理装置Ｘに備えられている。これによって、画像処理装置Ｘは、入力された画像データに領域分離処理を施し、画像データを構成する各ブロックの領域を識別することで、領域の特性に応じた様々な画像処理、たとえばフィルタ処理、ガンマ処理、中間調処理を施すことができる。これによって、高画質な画像を得ることができ、画像形成装置Ｙは、画像処理装置Ｘによって画像処理が施された画像データから高画質な画像を形成することができる。
【０１３４】
画像入力装置６０は、入力させる画像を読み取って画像データに変換するものであり、ＣＣＤ（Ｃｈａｒｇｅ Ｃｏｕｐｌｅｄ Ｄｅｖｉｃｅ）などを備えたスキャナ部を含んで構成され、画像からの反射光像をＣＣＤにて読み取り、画素ごとにＲＧＢのアナログ信号に変換し、このアナログ信号を画像処理装置Ｘに入力する。
【０１３５】
画像出力装置６１は、画像処理装置Ｘによって画像処理された画像データに基づいて、紙などの記録媒体上に画像を形成する。画像形成は、たとえば電子写真方式やインクジェット方式により行われる。
【０１３６】
次に、画像処理装置Ｘについて説明する。画像処理装置Ｘは、アナログ／デジタル（以後「Ａ／Ｄ」と略称する）変換部５０、シェーディング補正部５１、入力階調補正部５２、領域分離処理部Ａ、色補正部５３、墨生成下色除去処理部５４、空間フィルタ処理部５５、中間調出力階調処理部５６から構成される。
【０１３７】
画像入力装置６０から入力されたＲＧＢのアナログ信号の画像データは、Ａ／Ｄ変換部５０、シェーディング補正部５１、入力階調補正部５２、領域分離処理部Ａ、色補正部５３、墨生成下色除去処理部５４、空間フィルタ処理部５５、中間調出力階調処理部５６の順で処理され、ＣＭＹＫ（Ｃ：シアン、Ｍ：マゼンタ、Ｙ：イエロー、Ｋ：ブラック）のデジタルカラー信号として、画像出力装置６２へ出力される。
【０１３８】
Ａ／Ｄ変換部５０は、画像入力装置６０から入力されるＲＧＢのアナログ信号をデジタル信号に変換して、シェーディング補正部５１に出力する。シェーディング補正部５１は、Ａ／Ｄ変換部５０から入力されたＲＧＢのデジタル信号に対して、画像入力装置６０の照明系、結像系、撮像系で生じる各種の歪みを取り除く処理を施し、ＲＧＢの反射率信号として出力する。
【０１３９】
入力階調補正部５２は、シェーディング補正部５１にて各種の歪みが取り除かれたＲＧＢの反射率信号に対して、カラーバランスを整えるとともに、濃度など画像処理装置Ｘに採用されている画像処理システムに適した濃度信号に変換する。
【０１４０】
領域分離処理部Ａは、入力階調補正部５２から入力されたＲＧＢ信号の画像データに基づいて、画像データを、前述のようにブロックごとまたは画素ごとに文字、網点、写真、下地、不定の領域に分離していく領域分離処理を行う。領域分離処理部Ａは、処理の結果に基づき、各ブロックまたは各画素がいずれの領域に属しているかを示す領域識別信号を墨生成下色除去処理部５４、空間フィルタ処理部５５に出力するとともに、入力階調補正部５２から入力されたＲＧＢデジタル信号をそのまま色補正部５３に出力する。領域識別信号は、中間調出力階調処理部５６に出力されてもよい。
【０１４１】
色補正部５３は、ＲＧＢの濃度信号をＣＭＹ（Ｃ：シアン・Ｍ：マゼンタ・Ｙ：イエロー）の濃度信号に変換するとともに、色を忠実に再現するために、ＣＭＹの濃度信号に色補正処理を施す。ここでの色補正処理は、具体的には、不要吸収成分を含むＣＭＹのトナーやインクの分光特性に基づいた色濁りを、ＣＭＹの濃度信号から取り除く処理である。
【０１４２】
墨生成下色除去処理部５４は、領域分離処理部Ａから出力された領域識別信号を用い、領域ごとに、色補正部５３から出力されたＣＭＹの色信号に基づいて、黒（Ｋ）の色信号を生成する墨生成処理と、ＣＭＹの色信号に対して下色を除去する下色除去処理を施す。下色除去処理とは、ＣＭＹの色信号から墨生成処理で生成された黒の色信号を差し引いて新たなＣＭＹの色信号を得る処理である。これらの処理の結果、ＣＭＹの濃度信号は、ＣＭＹＫの色信号からなる画像データに変換される。
【０１４３】
空間フィルタ処理部５５は、墨生成下色除去処理部５４で得られたＣＭＹＫ画像データに対して、デジタルフィルタを用いた空間フィルタ処理を施す。当該処理も、領域分離処理部Ａから出力された領域識別信号を用いて、領域ごとに行われる。これによって画像の空間周波数特性が補正されるので、画像出力装置が出力する画像にぼやけ、または粒状性劣化を生じることを防止することができる。
【０１４４】
中間調出力階調処理部５６は、中間調生成処理を施す。中間調生成処理とは、画像を複数の画調に分割して階調を再現できるようにする処理であり、２値や多値のディザ法・誤差拡散法などを用いることができる。なお、中間調出力階調処理部５６は、画像データの濃度値を、画像出力装置６１の特性値である網点面積率に変換する処理を行ってもよい。そして、中間調出力階調処理部５６によって処理された濃度信号は、画像出力装置に与えられる。前述の画像処理装置Ｘの動作は、たとえば、画像処理装置Ｘが備えるＣＰＵ（Ｃｅｎｔｒａｌ Ｐｒｏｃｅｓｓｉｎｇ Ｕｎｉｔ）により制御される。
【０１４５】
本発明は、前述の領域分離処理方法を実行させるためのプログラムによって実現される。このプログラムは、画像処理方法を実行する装置の主メモリに直接格納されてもよいし、固定ディスク装置に挿入することによって読み取り可能となる記録媒体に格納されてもよい。これによって、領域分離処理方法を実行するプログラムを記録した記録媒体を持ち運び自在に提供することができる。
【０１４６】
なお、本実施の形態では、この記録媒体としては、マイクロコンピュータで処理が行われるためには、メモリ、たとえばＲＯＭ（Ｒｅａｄ Ｏｎｌｙ Ｍｅｍｏｒｙ）のようなものそのものがプログラムメディアであっても良いし、また、外部記憶装置としてプログラム読み取り装置が設けられ、そこに挿入することができる読み取り可能なプログラムメディアであってもよい。
【０１４７】
いずれの場合においても、格納されているプログラムはマイクロプロセッサがアクセスして実行させる構成であっても良いし、あるいは、いずれの場合もプログラムを読み出し、読み出されたプログラムは、マイクロコンピュータのプログラム記憶エリアにダウンロードされて、そのプログラムが実行される方式であってもよい。このダウンロード用のプログラムは予め本体装置に格納されているものとする。
【０１４８】
ここで、プログラムメディアは、本体と分離可能に構成される記録媒体であり、磁気テープやカセットテープなどのテープ系、フレキシブルディスクやハードディスクなどの磁気ディスクやＣＤ−ＲＯＭ（Ｃｏｍｐａｃｔ Ｄｉｓｋ Ｒｅａｄ Ｏｎｌｙ Ｍｅｍｏｒｙ）、ＭＯ（Ｍａｇｎｅｔｏ Ｏｐｔｉｃａｌ）ディスク、ＭＤ（Ｍｉｎｉ Ｄｉｓｋ）およびＤＶＤ（Ｄｉｇｉｔａｌ Ｖｅｒｓａｔｉｌｅ Ｄｉｓｋ）などの光ディスクのディスク系、メモリカードを含むＩＣ（Ｉｎｔｅｇｒａｔｅｄ Ｃｉｒｃｕｉｔｓ）カードおよび光カードなどのカード系、マスクＲＯＭ、ＥＰＲＯＭ（Ｅｒａｓａｂｌｅ Ｐｒｏｇｒａｍｍａｂｌｅ Ｒｅａｄ−Ｏｎｌｙ Ｍｅｍｏｒｙ）、ＥＥＰＲＯＭ（Ｅｌｅｃｔｒｉｃａｌｌｙ Ｅｒａｓａｂｌｅ Ｐｒｏｇｒａｍｍａｂｌｅ Ｒｅａｄ−Ｏｎｌｙ Ｍｅｍｏｒｙ）およびフラッシュＲＯＭなどによる半導体メモリを含めた固定的にプログラムを担持する媒体であってもよい。
【０１４９】
また、本実施形態が、インターネットを含む通信ネットワークを接続可能なシステム構成である場合、通信ネットワークからプログラムをダウンロードするように流動的にプログラムを担持する媒体であってもよい。なお、このように通信ネットワークからプログラムをダウンロードする場合には、そのダウンロード用のプログラムは予め本体装置に格納しておくか、あるいは別な記録媒体からインストールされるものであっても良い。
【０１５０】
前述の記録媒体は、デジタル画像形成装置やコンピュータシステムに備えられるプログラム読み取り装置により読み取られることにより、前述の領域分離処理方法が実行される。
【０１５１】
コンピュータシステムは、フラットベッドスキャナ・フィルムスキャナ・デジタルカメラなどの画像入力装置６０、所定のプログラムがロードされることにより画像処理方法など様々な処理が行われるコンピュータ、コンピュータの処理結果を表示するＣＲＴディスプレイ・液晶ディスプレイなどの画像表示装置、およびコンピュータの処理結果を紙などに出力するプリンタによって構成される。さらには、ネットワークを介してサーバーなどに接続するための通信手段としてのモデムなども備えられる。
【０１５２】
【発明の効果】
以上のように本発明によれば、エッジの方向という画像の構造的特徴に基づいて領域分離を行うので、空間周波数の成分が近似している網点領域と文字領域とを精度良く分離することができる。また、下地領域の判定は、検出した下地画素数を用いて行うので、容易に下地領域を分離することができる。また、写真領域の判定は、検出したエッジ画素数を用いて行うので、容易に写真領域を分離することができる。
【０１５３】
また本発明によれば、水平エッジ成分と垂直エッジ成分との差分を算出するだけでエッジの方向を決定できるので、計算量を削減することができ、精度を下げることなく領域分離処理の高速化を図ることができる。また、領域分離処理を行う回路規模を縮小することができる。
【０１５４】
また本発明によれば、注目画素のエッジ方向と同じエッジ方向を有する近傍エッジ画素数に基づいて特徴量を算出するので、画素ブロックの特徴量を容易に算出することができる。
【０１５５】
また本発明によれば、エッジ画素のエッジの方向を、その方向に応じて所定の値に置き換えて、注目画素のエッジ方向と同じエッジ方向を有する近傍エッジ画素数を算出するので、判断処理を行う必要がなく、計算量を削減することができ、領域分離処理の高速化を図ることができる。また、領域分離処理を行う回路規模を縮小することができる。
【０１５６】
また本発明によれば、上記の領域分離処理方法をコンピュータに実行させるための領域分離処理プログラムとして提供することができる。
【０１５７】
また本発明によれば、上記の領域分離処理方法をコンピュータに実行させるための領域分離処理プログラムを記録したコンピュータ読み取り可能な記録媒体として提供することができる。
【０１５８】
また本発明によれば、高精度に領域分離が行われ、領域に応じた処理が施された画像データを出力することができるので、高画質の画像を形成することができる。
【図面の簡単な説明】
【図１】文字画像と網点画像におけるエッジ方向の分布を示す図である。
【図２】領域分離処理を実行する領域分離処理部Ａの構成を示すブロック図である。
【図３】領域分離処理部Ａによる領域分離処理の手順を示すフローチャートである。
【図４】図２に示すブロック判定部３の構成を示すブロック図である。
【図５】ブロック判定部３によるブロック判定処理の手順を示すフローチャートである。
【図６】下地画素検出処理部１０による下地画素検出処理の詳細な手順を示すフローチャートである。
【図７】水平エッジ成分算出処理部１１１による水平エッジ成分算出処理の手順を示すフローチャートである。
【図８】垂直エッジ成分算出処理部１１２による垂直エッジ成分算出処理の手順を示すフローチャートである。
【図９】エッジ方向分類処理部１２によるエッジ方向分類処理の詳細な手順を示すフローチャートである。
【図１０】角度によるエッジ方向の分類方法を表す図である。
【図１１】エッジ方向分類処理の他の手順を示すフローチャートである。
【図１２】方向相関算出処理部１３による方向相関算出処理の詳細な手順を示すフローチャートである。
【図１３】内積を利用して方向相関算出処理を行う場合の処理手順を示すフローチャートである。
【図１４】判定処理の詳細な手順を示すフローチャートである。
【図１５】画像形成装置Ｙの構成を示すブロック図である。
【符号の説明】
１ ブロック分割部
２ Ｌ^＊ａ^＊ｂ^＊変換部
３ ブロック判定部
１０ 下地画素検出処理部
１１ エッジ成分算出処理部
１２ エッジ方向分類処理部
１３ 方向相関算出処理部
１４ 判定処理部
５０ アナログ／デジタル（Ａ／Ｄ）変換部
５１ シェーディング補正部
５２ 入力階調補正部
５３ 色補正部
５４ 墨生成下色除去処理部
５５ 空間フィルタ処理部
５６ 中間調出力階調処理部
６０ 画像入力装置
６１ 画像出力装置
Ａ 領域分離処理部
Ｘ 画像処理装置
Ｙ 画像形成装置[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to an image forming apparatus such as a digital copying machine, in which an image data is separated into a base area, a character area, a halftone area, and a photograph area in order to appropriately perform processing such as filter processing and halftone generation processing. The present invention relates to a separation processing method, an area separation processing program, and a recording medium on which the program is recorded. Furthermore, the present invention relates to an image processing apparatus provided with an area separation processing unit using the above-described area separation method, and an image forming apparatus provided with the image processing apparatus and outputting an image using image data obtained by area separation.
[0002]
[Prior art]
In general digital copiers and the like, in order to output a high-quality image, each pixel is separated into a base area, a character area, a halftone area, and a photographic paper photographic area (photo area) with respect to input image data. And image processing such as filter processing, gamma processing, and halftone processing according to the characteristics of each area.
[0003]
For example, Patent Document 1 discloses an image processing apparatus that performs a wavelet transform process on a plurality of layers to input image data to obtain a plurality of different frequency components, and performs a region separation process based on the obtained frequency components. ing.
[0004]
The wavelet transform is a type of frequency conversion, and is a technique for performing encoding using the redundancy of frequency-converted image data. The different frequency components refer to a DC component, an AC component in the main scanning direction, an AC component in the sub-scanning direction, and an AC component in an oblique direction. When the wavelet transform processing is performed on the image data, the image can be decomposed into a low-frequency component and a high-frequency component, and a wavelet decomposition coefficient can be obtained. Since various information such as color, texture, and shape information can be obtained from the wavelet decomposition coefficient, the wavelet decomposition coefficient can be used as a feature amount of image data.
[0005]
In the image processing apparatus of Patent Document 1, first, input image data is divided into blocks each including a plurality of pixel data. Then, wavelet transform is performed on the image data in the block to extract a predetermined frequency component, and a first threshold value corresponding to the frequency component is selected. The absolute value of a plurality of coefficients constituting the extracted frequency component is compared with the selected first threshold. The block is separated into a character area, a halftone area, and a continuous tone area by counting the number of coefficients whose absolute value exceeds a first threshold value and comparing the counted number with a predetermined second threshold value. .
[0006]
[Patent Document 1]
JP-A-11-220622
(Page 4-11, Figures 1, 2, 7, 19)
[0007]
[Problems to be solved by the invention]
In the case where region segmentation is performed by using the wavelet transform, in the past, a region was segmented after performing a wavelet inverse transform and restoring the original image data. According to the image forming apparatus, since the signal decomposed into different frequency components after the wavelet transform can be separated into regions without performing the inverse wavelet transform, extra labor and time can be saved.
[0008]
However, since the above technique focuses on the difference between the frequency components of the respective regions, the regions are separated from each other, so that regions having similar frequency distributions, such as a halftone dot region having a small number of lines and a character region including small characters, are used. Is very difficult to separate.
[0009]
Further, in the above technique, in order to improve the segmentation accuracy, it is necessary to perform a wavelet transform of two or more layers, the amount of calculation becomes enormous, and there is a problem from the viewpoint of processing speed. Further, in order to perform the region separation, it is necessary to process a large number of coefficient signals generated by the wavelet transform of a plurality of layers, and the amount of calculation becomes enormous. However, there is a problem that the processing cost is increased accordingly.
[0010]
An object of the present invention is to provide an area separation processing method, an area separation processing program, and a recording medium on which the area separation processing method and the high-quality image can be accurately and rapidly performed. An object of the present invention is to provide an image processing apparatus that can generate the image data and an image forming apparatus including the image processing apparatus.
[0011]
[Means for Solving the Problems]
According to the present invention, image data representing an image composed of a plurality of pixels is input, and based on the input image data, a pixel block composed of a plurality of pixels or a target pixel in the pixel block includes a character area and a halftone dot area In an area separation processing method of determining which of a plurality of areas belongs to and performing area separation of image data,
A dividing step of dividing the image data into pixel blocks each having a predetermined number of pixels;
A component calculation step of calculating an edge component of each pixel in the pixel block,
A direction determining step of determining a direction of an edge formed by each pixel based on the calculated edge component;
A feature amount calculating step of calculating a feature amount of the pixel block based on the correlation between the direction of the edge of each pixel in the pixel block and the direction of the edge of a neighboring pixel based on the determined direction of the edge;
Determining whether the pixel block or the pixel of interest in the pixel block belongs to the character area or the halftone dot area based on the calculated feature amount. It is.
[0012]
The direction of the edge formed by each pixel in the pixel block indicates the characteristic of the pixel block or the pixel of interest in the pixel block. A pixel block or a target pixel in a pixel block including many pixels having different edge directions belongs to a dot region, and a pixel block or a target pixel in a pixel block including many edge pixels having the same edge direction is a character. Belongs to the area. The present invention focuses on these, and performs the region separation processing.
[0013]
According to the present invention, first, in a dividing step, input image data is divided into pixel blocks each having a predetermined number of pixels.
[0014]
In the component calculation step, an edge component of each pixel in the pixel block is calculated. The edge component is a value that is referred to when determining whether the target pixel is a pixel forming an edge, and for example, a gradation difference between the target pixel and a neighboring pixel is used. The edge component includes a horizontal edge component and a vertical edge component. For example, each component is calculated using an edge component extraction filter or the like.
[0015]
In the direction determining step, the direction of the edge formed by each pixel is determined based on the calculated edge component. If the horizontal edge component is greater than the vertical edge component, the edge direction is horizontal, and if the vertical edge component is greater than the horizontal edge component, the edge direction is vertical. If the difference between the horizontal edge component and the vertical edge component is small, the edge direction is an oblique direction. Therefore, the edge direction is determined by comparing the horizontal edge component and the vertical edge component.
[0016]
In the feature amount calculating step, the feature amount of the pixel block is calculated based on the correlation between the direction of the edge of each pixel in the pixel block and the direction of the edge of its neighboring pixels. Here, the correlation indicates how much the direction of the edge of each pixel and the direction of the edge of a neighboring pixel are the same or different, that is, the ratio of neighboring pixels having the same edge direction among neighboring pixels.
[0017]
As described above, the direction of the edge of each pixel in the pixel block represents the characteristic of the pixel block or the target pixel in the pixel block. The number of pixels having an edge direction different from the neighboring pixels included in the pixel block is calculated as a feature amount.
[0018]
In the determining step, it is determined whether or not the pixel block or the target pixel in the pixel block belongs to the character area or the halftone dot area based on the calculated feature amount. For example, when the feature amount is the above-described feature amount, the feature amount is compared with a predetermined threshold. If the feature value is equal to or larger than the threshold value, that is, if many pixels having an edge direction different from the neighboring pixels are included, it is determined that the pixel block or the target pixel in the pixel block belongs to the halftone dot region. If the feature value is smaller than the threshold value, that is, if many pixels having the same edge direction as the neighboring pixels are included, it is determined that the pixel block or the target pixel in the pixel block belongs to the character area.
[0019]
Thus, since the area is separated based on the structural feature of the image, that is, the direction of the edge, it is possible to accurately separate the halftone dot area and the character area whose spatial frequency components are similar.
[0020]
Further, according to the present invention, image data representing an image composed of a plurality of pixels is input, and based on the input image data, a pixel block composed of a plurality of pixels or a pixel of interest in the pixel block is a character region, a halftone dot region and In an area separation processing method of determining which of a plurality of areas including a photograph area belongs, and performing area separation of image data,
A dividing step of dividing the image data into pixel blocks each having a predetermined number of pixels;
A component calculation step of calculating an edge component of each pixel in the pixel block,
An edge pixel counting step of detecting an edge pixel in the pixel block based on the edge component and counting the number of edge pixels;
A direction determining step of determining a direction of an edge formed by each edge pixel based on the edge component,
A feature value calculating step of calculating a feature value of the pixel block based on the correlation between the direction of the edge of each edge pixel in the pixel block and the direction of the edge of its neighboring edge pixels based on the determined direction of the edge;
Based on the counted number of edge pixels, it is determined whether or not the pixel of interest in the pixel block or the pixel block belongs to a photographic area, and based on the calculated feature amount, A determining step of determining whether or not a pixel belongs to the character area or the halftone dot area.
[0021]
According to the present invention, first, in the dividing step, input image data is divided into pixel blocks each having a predetermined number of pixels.
[0022]
In the component calculation step, an edge component of each pixel in the pixel block is calculated. The edge component includes a horizontal edge component and a vertical edge component. For example, each component is calculated using an edge component extraction filter or the like.
[0023]
When the edge component is calculated, in the edge pixel counting step, an edge pixel in the pixel block is detected based on the edge component, and the number of edge pixels is counted. Since an edge pixel has a large edge component, a pixel whose edge component is larger than a predetermined threshold is detected as an edge pixel.
[0024]
In the direction determining step, the direction of the edge formed by each edge pixel is determined based on the calculated edge component. If the horizontal edge component is greater than the vertical edge component, the edge direction is horizontal, and if the vertical edge component is greater than the horizontal edge component, the edge direction is vertical. If the difference between the horizontal edge component and the vertical edge component is small, the direction of the edge is oblique. Therefore, the direction of the edge is determined by comparing the horizontal edge component and the vertical edge component.
[0025]
In the feature amount calculation step, the feature amount of the pixel block is calculated based on the correlation between the edge direction of each edge pixel in the pixel block and the edge direction of the neighboring pixels based on the determined edge direction. As described above, the edge direction of the edge pixel in the pixel block indicates the characteristic of the pixel block or the target pixel in the pixel block. The number of edge pixels having an edge direction different from that of the neighboring edge pixels included in the pixel block is calculated as a feature amount.
[0026]
In the determination step, for example, the counted number of edge pixels is compared with a predetermined threshold value, and if it is smaller than the threshold value, it is determined that the pixel block or the target pixel in the pixel block belongs to the photograph area. When it is determined that the pixel does not belong to the photograph area, it is determined whether or not the pixel block or the target pixel in the pixel block belongs to the character area or the halftone area based on the calculated feature amount. For example, when the feature amount is the above-described feature amount, the feature amount is compared with a predetermined threshold. When the feature value is equal to or larger than the threshold value, that is, when many edge pixels having an edge direction different from the neighboring edge pixels are included, it is determined that the pixel block or the pixel of interest in the pixel block belongs to the dot area. When the feature value is smaller than the threshold value, that is, when many edge pixels having the same edge direction as the neighboring edge pixels are included, it is determined that the pixel block or the target pixel in the pixel block belongs to the character area.
[0027]
Thus, since the area is separated based on the structural feature of the image, that is, the direction of the edge, it is possible to accurately separate the halftone dot area and the character area whose spatial frequency components are similar. In addition, since the determination of the photograph area is performed using the detected number of edge pixels, the photograph area can be easily separated.
[0028]
Further, according to the present invention, image data indicating an image composed of a plurality of pixels is input, and based on the input image data, a pixel block composed of a plurality of pixels or a target pixel in the pixel block is a character area, a halftone area, In an area separation processing method of determining which of a plurality of areas including a photograph area and a base area belongs, and performing area separation of image data,
A dividing step of dividing the image data into pixel blocks each having a predetermined number of pixels;
A base pixel counting step of detecting a base pixel in the pixel block based on the tone value of the pixel and counting the number of base pixels;
A component calculation step of calculating an edge component of each pixel in the pixel block,
An edge pixel counting step of detecting an edge pixel in the pixel block based on the edge component and counting the number of edge pixels;
A direction determining step of determining a direction of an edge formed by each edge pixel based on the edge component,
A feature value calculating step of calculating a feature value of the pixel block based on the correlation between the direction of the edge of each edge pixel in the pixel block and the direction of the edge of its neighboring edge pixels based on the determined direction of the edge;
It is determined whether or not the pixel of interest in the pixel block or the pixel block belongs to the background area based on the counted number of background pixels, and the pixel of interest in the pixel block or pixel block is determined based on the counted number of edge pixels. Determining whether the pixel block belongs to a photograph area, and determining whether the pixel of interest in the pixel block or the pixel block belongs to the character area or the halftone dot area based on the calculated feature amount; and Is a region separation processing method.
[0029]
According to the present invention, first, in the dividing step, input image data is divided into pixel blocks each having a predetermined number of pixels.
[0030]
In the background pixel counting step, the background pixels in the pixel block are detected based on the gradation values of the pixels, and the number of background pixels is counted. Since the background area is often white, the gradation value of the background pixel is large. Therefore, a pixel whose tone value is larger than a predetermined threshold value is detected as a base pixel.
[0031]
In the component calculation step, an edge component of each pixel in the pixel block is calculated. The edge component includes a horizontal edge component and a vertical edge component. For example, each component is calculated using an edge component extraction filter or the like.
[0032]
When the edge component is calculated, in the edge pixel counting step, an edge pixel in the pixel block is detected based on the edge component, and the number of edge pixels is counted. Since an edge pixel has a large edge component, a pixel whose edge component is larger than a predetermined threshold is detected as an edge pixel.
[0033]
In the direction determining step, the direction of the edge formed by each edge pixel is determined based on the calculated edge component. If the horizontal edge component is greater than the vertical edge component, the edge direction is horizontal, and if the vertical edge component is greater than the horizontal edge component, the edge direction is vertical. If the difference between the horizontal edge component and the vertical edge component is small, the direction of the edge is oblique. Therefore, the direction of the edge is determined by comparing the horizontal edge component and the vertical edge component.
[0034]
In the feature amount calculating step, the feature amount of the pixel block is calculated based on the correlation between the edge direction of each edge pixel in the pixel block and the edge direction of its neighboring pixels. As described above, the edge direction of the edge pixel in the pixel block indicates the characteristic of the pixel block or the target pixel in the pixel block. The number of edge pixels having an edge direction different from that of the neighboring edge pixels included in the pixel block is calculated as a feature amount.
[0035]
Finally, in the determination step, for example, the counted number of background pixels is compared with a predetermined threshold value, and if it is larger than the threshold value, it is determined that the pixel block or the target pixel in the pixel block belongs to the background region. When it is determined that the pixel does not belong to the background area, for example, the counted number of edge pixels is compared with a predetermined threshold. If the number is smaller than the threshold, it is determined that the pixel block or the pixel of interest in the pixel block belongs to the photograph area. When it is determined that the pixel does not belong to the photograph area, it is determined whether or not the pixel block or the target pixel in the pixel block belongs to the character area or the halftone area based on the calculated feature amount. For example, when the feature amount is the above-described feature amount, the feature amount is compared with a predetermined threshold. When the feature value is equal to or larger than the threshold value, that is, when many edge pixels having an edge direction different from the neighboring edge pixels are included, it is determined that the pixel block or the pixel of interest in the pixel block belongs to the dot area. When the feature value is smaller than the threshold value, that is, when many edge pixels having the same edge direction as the neighboring edge pixels are included, it is determined that the pixel block or the target pixel in the pixel block belongs to the character area.
[0036]
Thus, since the region is separated based on the structural feature of the image, that is, the direction of the edge, it is possible to accurately separate the halftone dot region and the character region in which the spatial frequency components are approximated. In addition, since the background area is determined using the detected number of background pixels, the background area can be easily separated. In addition, since the determination of the photograph area is performed using the detected number of edge pixels, the photograph area can be easily separated.
[0037]
Further, in the present invention, the component calculating step calculates a horizontal edge component and a vertical edge component as edge components,
The direction determining step may determine an edge direction from a difference between the horizontal edge component and the vertical edge component.
[0038]
According to the present invention, a horizontal edge component and a vertical edge component are calculated as edge components, and the direction of the edge is determined from the difference between these components.
[0039]
As described above, the direction of the edge is determined only by calculating the difference between the horizontal edge component and the vertical edge component. For example, the angle of the edge is calculated from the arc tangent of the ratio between the horizontal edge component and the vertical edge component. As compared with the case of performing the calculation, the amount of calculation can be reduced, and the speed of the region separation processing can be increased without lowering the accuracy. Further, the circuit scale for performing the region separation processing can be reduced.
[0040]
Further, in the feature amount calculating step, when each edge pixel in the pixel block is set as a target pixel, the number of neighboring edge pixels having the same edge direction as the edge direction of the target pixel among the neighboring edge pixels of the target pixel Is counted for each pixel of interest,
The number of pixels of the target pixel in which the counted number of neighboring edge pixels is smaller than a predetermined value is calculated as a feature amount.
[0041]
According to the present invention, in the feature amount calculation step, first, an edge pixel in a pixel block is set as a target pixel, and among neighboring edge pixels of the target pixel, the number of neighboring edge pixels having the same edge direction as the edge direction of the target pixel ( The number of neighboring edge pixels in the same direction) is counted. The target pixel is sequentially changed to another edge pixel in the pixel block, and the number of edge pixels in the same direction in the same direction is counted for each target pixel. The number of neighboring edge pixels in the same direction counted for each pixel of interest is compared with a predetermined value, for example, １ ／ of the number of neighboring edge pixels, and the number of pixels of the pixel of interest smaller than this is calculated.
[0042]
As described above, since the feature amount is calculated based on the number of edge pixels in the same direction, the feature amount of the pixel block can be easily calculated.
[0043]
Further, in the present invention, in the feature amount calculating step, the direction of the edge of the edge pixel is replaced with a predetermined value according to the direction,
The number of neighboring edge pixels having the same edge direction as the target pixel is counted using the replaced value.
[0044]
According to the present invention, the direction of the edge of the edge pixel is replaced with a predetermined value according to the direction. As the predetermined value, for example, an arbitrary vector is used. Each vector is set such that the inner product of the vectors is "1" when the edges are in the same direction, and "0" when the edges are in different directions. With this setting, when counting neighboring edge pixels in the same direction, it is only necessary to find the sum of the inner products of the vector indicating the edge direction of the target pixel and the vector indicating the edge direction of the neighboring edge pixel. It is not necessary to determine whether the edge direction is the same as the edge direction of the neighboring edge pixels.
[0045]
As described above, since the determination process is not performed, the amount of calculation can be reduced, and the speed of the region separation process can be increased. Further, the circuit scale for performing the region separation processing can be reduced.
[0046]
Further, the present invention is an area separation processing program for causing a computer to execute the above area separation processing method.
[0047]
According to the present invention, an area separation processing program for causing a computer to execute the above area separation processing method can be provided.
[0048]
According to the present invention, there is provided a computer-readable recording medium recording an area separation processing program for causing a computer to execute the above-described area separation processing method.
[0049]
According to the present invention, it is possible to provide a computer-readable recording medium that records an area separation processing program for causing a computer to execute the above-described area separation processing method.
[0050]
Further, according to the present invention, image data representing an image composed of a plurality of pixels is input, and based on the input image data, a pixel block composed of a plurality of pixels or a pixel of interest in the pixel block defines a character region and a halftone dot region. In an image processing apparatus including an area separation processing unit that determines which of a plurality of areas including the image data belongs to and performs area separation of image data,
Dividing means for dividing the image data into pixel blocks each having a predetermined number of pixels;
Component calculating means for calculating an edge component of each pixel in the pixel block,
Direction determining means for determining the direction of the edge formed by each pixel based on the calculated edge component,
A feature amount calculating unit configured to calculate a feature amount of the pixel block based on a correlation between the direction of the edge of each pixel in the pixel block and the direction of the edge of a neighboring pixel based on the determined direction of the edge;
A determination unit configured to determine whether the pixel of interest or the pixel of interest in the pixel block belongs to the character region or the halftone dot region based on the calculated feature amount. is there.
[0051]
According to the present invention, first, the dividing means divides the input image data into pixel blocks each having a predetermined number of pixels.
[0052]
The component calculation means calculates an edge component of each pixel in the pixel block. The edge component includes a horizontal edge component and a vertical edge component. For example, each component is calculated using an edge component extraction filter or the like.
[0053]
Based on the calculated edge component, the direction determining means determines the direction of the edge formed by each pixel. If the horizontal edge component is greater than the vertical edge component, the edge direction is horizontal, and if the vertical edge component is greater than the horizontal edge component, the edge direction is vertical. If the difference between the horizontal edge component and the vertical edge component is small, the direction of the edge is oblique. Therefore, the direction of the edge is determined by comparing the horizontal edge component and the vertical edge component.
[0054]
The characteristic amount calculating means calculates the characteristic amount of the pixel block based on the correlation between the edge direction of each pixel in the pixel block and the edge directions of neighboring pixels based on the determined edge direction. As described above, the edge direction of the edge pixel in the pixel block indicates the characteristic of the pixel block or the target pixel in the pixel block. The number of edge pixels having an edge direction different from that of the neighboring edge pixels included in the pixel block is calculated as a feature amount.
[0055]
Based on the calculated feature amount, the determination unit determines whether the pixel block or the pixel of interest in the pixel block belongs to a character area or a halftone dot area. For example, when the feature amount is the above-described feature amount, the feature amount is compared with a predetermined threshold. When the feature value is equal to or larger than the threshold value, that is, when many edge pixels having an edge direction different from the neighboring edge pixels are included, it is determined that the pixel block or the pixel of interest in the pixel block belongs to the dot area. When the feature value is smaller than the threshold value, that is, when many edge pixels having the same edge direction as the neighboring edge pixels are included, it is determined that the pixel block or the target pixel in the pixel block belongs to the character area.
[0056]
Thus, since the region is separated based on the structural feature of the image, that is, the direction of the edge, it is possible to accurately separate the halftone dot region and the character region in which the spatial frequency components are approximated.
[0057]
The present invention also provides the image processing device described above,
An image forming apparatus comprising: an image output device that outputs image data processed by the image processing device.
[0058]
According to the invention, the image data processed by the image processing device is output from the image output device.
[0059]
As a result, high-accuracy region separation is performed, and image data that has been subjected to processing according to the region can be output, so that a high-quality image can be formed.
[0060]
BEST MODE FOR CARRYING OUT THE INVENTION
The present invention relates to a region separation process. The region separation processing is defined as any one of a halftone dot region, a character region, a photographic paper photograph region (photo region), a background region, and an undefined region where it cannot be determined which of these four regions belongs to each pixel of the input image data. This is the process of separating crab. In this way, by separating the input image data into each area, it is possible to perform image processing such as filter processing, gamma processing, and halftone processing according to the characteristics of each area, and output a high-quality image. Can be.
[0061]
An area separation processing method according to an embodiment of the present invention will be described. In the present embodiment, first, the input image data is divided into pixel blocks each having a rectangle of an arbitrary size. Then, the base region is separated from the plurality of types of regions. The background region is also called a white background region, and is a background portion where the brightness of an image is equal to or higher than a certain value. Therefore, the region can be separated based on the brightness. After separating a block having a certain lightness or more as a background area, the horizontal direction, the vertical direction, and the diagonal edge are detected for each block in three directions, and the total number of edge pixels in the pixel block and the correlation in the edge direction are detected. And the like are calculated based on. The edge direction of a pixel refers to the direction of the edge formed by the pixel.
[0062]
The photographic paper photographic area is an image having a halftone density, and the number of edges present in the image is smaller than that of other areas. Can be separated.
[0063]
In addition, there is a structural difference between the character area and the halftone dot area, which appears in the appearance of edges. FIG. 1 is a diagram showing a distribution of edge directions in a character image and a halftone image. FIG. 1A illustrates a character image, and FIG. 1B illustrates a halftone image. In both the character image and the halftone image, the input image is shown on the upper left and the edge image is shown on the upper right. The lower edge shows a horizontal edge image and a vertical edge obtained by further dividing the upper edge image into edges in three directions. The image and the oblique edge image are shown in the order of left, center, and right.
[0064]
As shown in the lower part of FIG. 1A, among the three directions of edges that appear in the character image, the positions where the edges in different directions appear are greatly different, and the proportion of the edges in different directions appearing in the predetermined area is low. . On the other hand, as shown in the lower part of FIG. 1B, among the three directions of edges appearing in the halftone dot image, the positions where the edges of different directions appear are not so different, and the edges of different directions are within the predetermined area. The rate of appearance is high.
[0065]
Thus, the character area has a structure in which many edge pixels in the same direction are included, and the halftone dot area has a structure in which many edge pixels in different directions are included. Based on this, both areas can be separated.
[0066]
FIG. 2 is a block diagram illustrating a configuration of the area separation processing unit A that performs the area separation processing. The area separation processing unit A includes a block division unit 1, L ^* a ^* b ^* It comprises a converter 2 and a block determiner 3. The block division unit 1 refers to input image data composed of RGB signals (R: red, G: green, B: blue) as a block of pixels composed of m × n pixels (hereinafter, referred to as a “determination block”). , M, and n are arbitrary natural numbers.) ^* a ^* b ^* Output to the converter 2. L ^* a ^* b ^* The conversion unit 2 converts the RGB signals of the pixels of each determination block divided by the block division unit 1 into CIE1976L ^* a ^* b ^* Signal (CIE: Commission Internationale de l'Eclairage: International Commission on Illumination. L ^* : Lightness, a ^* , B ^* : Chromaticity), and ^* a ^* b ^* L of the signal ^* The signal is output to the block determination unit 3. Color space conversion is CIE1976L ^* a ^* b ^* This is done by using the definition formula of
[0067]
The block determining unit 3 converts the converted CIE1976L ^* a ^* b ^* L of the signal ^* Using the signal, each determination block is classified into a halftone block composed of a halftone area, a character block composed of a character area, a photographic block composed of a photographic area, a base block composed of a background area, and an undefined block which is an undefined area. Output the identification signal. In the present embodiment, the CIE1976L ^* a ^* b ^* L of the signal ^* The image data is subjected to the area separation processing using the signal, and the signal used for the area separation processing is L. ^* The luminance value is not limited to the signal, and may be a luminance value represented by the equation (1), or may be performed using an RGB signal as it is.
Y _j = 0.30R _j + 0.59G _j + 0.11B _j … (1)
[0068]
Y here _j Represents the luminance value of each pixel, and R _j , G _j , B _j Represents the color component of each pixel.
[0069]
The reading resolution of the input image data needs to be at least twice the printing resolution in order to avoid aliasing (such as interference fringes not present in the original), and the higher the reading resolution, the higher the determination accuracy.
[0070]
In addition, the larger the size of the determination block is, the higher the determination accuracy is. For example, a size of m × n = 7 × 7 pixels or more is desirable.
[0071]
FIG. 3 is a flowchart illustrating a procedure of the area separation processing by the area separation processing unit A. First, in step a1, the block dividing unit 1 performs a dividing process of dividing input image data into a determination block of pixels of m × n pixels (m and n are arbitrary natural numbers). In step a2, L ^* a ^* b ^* The conversion unit 2 converts the RGB signals of each decision block to L ^* a ^* b ^* Convert to a signal. In step a3, a block determination process is performed in which the block determination unit 3 determines whether each determination block belongs to a halftone dot, a character, a photograph, a background, or an undefined area. Then, in step a4, it is determined whether or not each process has been completed for all determination blocks in the image data. If the processing has been completed, the area separation processing ends, and if not, the processing returns to step a3.
[0072]
FIG. 4 is a block diagram showing a configuration of the block determination unit 3 shown in FIG. The block determination unit 3 includes a background pixel detection processing unit 10, an edge component calculation processing unit 11, an edge direction classification processing unit 12, a direction correlation calculation processing unit 13, and a determination processing unit 14. Further, the edge component calculation processing section 11 includes a horizontal edge component calculation processing section 111 and a vertical edge component calculation processing section 112. The edge component calculation processing unit 11 forms an edge extraction unit, the edge direction classification processing unit 12 forms an edge direction determination unit, the direction correlation calculation processing unit 13 forms a direction correlation detection unit, and the determination processing unit 14 determines Configure means.
[0073]
The background pixel detection processing unit 10 detects a background pixel and outputs a detection result to the determination processing unit 14 as a background pixel signal. The edge component calculation processing unit 11 calculates an edge component, and outputs the calculation result to the edge direction classification processing unit 12 as an edge component signal. Here, the edge component indicates a gradation difference between a pixel to be processed (hereinafter, referred to as a “target pixel”) in a determination block and a neighboring pixel, and indicates whether the target pixel is a pixel forming an edge. Is a value that is referred to when determining. The edge component can be calculated by performing a filtering process on a predetermined pixel data group including the pixel data of the target pixel. The edge component is calculated according to the direction of the edge. In the present embodiment, the horizontal and vertical edge components are calculated. The horizontal edge component calculation processing unit 111 calculates a horizontal edge component, and the vertical edge component calculation processing unit 112 calculates a vertical edge component.
[0074]
The edge direction classification processing unit 12 converts the pixels in the determination block into non-edge pixels, horizontal edge pixels, vertical edge pixels, and oblique edge pixels based on the horizontal and vertical edge component signals output from the edge component calculation processing unit 11. Classification is performed, and the classification result is output to the direction correlation calculation processing unit 13 as an edge direction signal. The direction correlation calculation processing unit 13 calculates a feature amount based on the correlation in the edge direction based on the edge direction signal, and outputs the calculation result to the determination processing unit 14 as a direction correlation signal. The determination processing unit 14 determines to which region the determination block belongs based on the background pixel signal and the direction correlation signal. Then, the determination result is output as an area identification signal.
[0075]
FIG. 5 is a flowchart illustrating a procedure of a block determination process performed by the block determination unit 3.
[0076]
First, in step b1, the background pixel detection processing unit 10 of the block determination unit 3 performs background pixel detection processing. In step b2, the horizontal edge component calculation processing section 111 performs a horizontal edge component calculation process. In step b3, the vertical edge component calculation processing section 112 performs a vertical edge component calculation process.
[0077]
In step b4, the edge direction classification processing unit 12 performs edge direction classification processing of classifying pixels in the determination block into non-edge pixels, horizontal edge pixels, vertical edge pixels, and oblique edge pixels. In step b5, the direction correlation calculation processing unit 13 performs a direction correlation calculation process of calculating a feature amount based on the correlation in the edge direction. In step b6, the determination processing unit 14 performs a determination process of determining whether the determination block of the input image data is a base block, a photograph block, a character block, a halftone block, or an undefined block. The determination processing ends.
[0078]
Next, background pixel detection processing (step b1 shown in FIG. 5), horizontal and vertical edge component calculation processing (steps b2 and b3 shown in FIG. 5), edge direction classification processing (step shown in FIG. 5) which constitutes the block determination processing b4), detailed procedures of the directional correlation calculation process (step b5 shown in FIG. 5) and the determination process (step b6 shown in FIG. 5) will be described.
[0079]
FIG. 6 is a flowchart illustrating a detailed procedure of the background pixel detection processing by the background pixel detection processing unit 10. The base pixel is pixel data L representing lightness. ^* Has a value equal to or greater than a certain value. Here, the brightness L of the input image data is determined for each pixel in the determination block. ^* Is compared with a predetermined threshold value, and how many background pixels exist in the determination block is calculated. This processing is performed in order from the leftmost pixel to the rightmost pixel in the upper stage in the determination block, and is repeated from each pixel in the uppermost stage to the lower right pixel in the determination block.
[0080]
First, in step c1, the brightness L of the pixel data in the determination block is ^* Is greater than or equal to a predetermined threshold T1. If it is larger than the threshold T1, the process proceeds to step c2, and if it is smaller than the threshold T1, the process proceeds to step c3. In step c2, 1 is added to the number of base pixels Cnt_white. In step c3, it is determined whether the processing has been completed for all pixels in the determination block. If the processing has been completed, the background pixel detection processing has been completed. If the processing has not been completed, the processing returns to step c1. By this processing, the number of base pixels Cnt_white in the determination block is calculated.
[0081]
Next, the horizontal edge component calculation processing (step b2 shown in FIG. 5) will be described. FIG. 7 is a flowchart illustrating a procedure of the horizontal edge component calculation processing by the horizontal edge component calculation processing unit 111. Here, for each pixel constituting the determination block, the horizontal edge component of the pixel data is calculated as a numerical value, and the calculated value is used in edge direction classification processing described later.
[0082]
In step d1, a horizontal edge component is calculated based on the following equation (2). Here, the horizontal edge component is represented by hor, the edge component calculation function is represented by Edge, and the pixel data of the pixel group to be processed is represented by OriginalL. The pixel group to be processed is, for example, a 3 × 3 pixel group, and includes a pixel of interest at the center of the 3 × 3 pixel group and eight neighboring pixels around the pixel of interest.
hor = Edge (filter_hor, OriginalL) (2)
[0083]
Note that filter_hor is a filter for extracting a horizontal edge component. As this filter, a high-pass filter (High Pass Filter: HPF) that passes high-frequency components may be used, or a low-pass filter (Low Pass Filter: LPF) that passes low-frequency components may be used. However, the content of the operation differs as follows depending on the type of the filter.
[0084]
In the case of the HPF, the convolution operation of the pixel data OriginalL of the target pixel group and the filter filter_hor is performed as shown in Expression (3) to calculate the horizontal edge component of the target pixel.
[0085]
(Equation 1)

[0086]
In the case of the LPF, the convolution operation of the pixel data OriginalL of the pixel group to be processed and the filter filter_hor is performed as shown in Expression (4), and this is subtracted from the pixel data OriginalL of the original pixel group. By doing so, it is possible to extract only the horizontal edge component of the target pixel while suppressing the influence of noise.
[0087]
(Equation 2)

[0088]
The filter may have any shape as long as it has either high-pass or low-pass properties. For example, a Laplacian filter can be used.
[0089]
In addition, the filtering process for the pixel located at the end of the determination block is performed by interleaving, for example, by interpolating pixel data of an adjacent pixel in the determination block as pixel data outside the determination block.
[0090]
Next, the vertical edge component calculation processing (step b3 shown in FIG. 5) will be described. FIG. 8 is a flowchart illustrating the procedure of the vertical edge component calculation processing by the vertical edge component calculation processing unit 112. Here, the vertical edge component of the pixel data is calculated as a numerical value for each pixel constituting the determination block, and the calculated value is used in an edge direction classification process described later. The basic processing is the same as the horizontal edge component processing.
[0091]
In step e1, the vertical edge component ver of the target pixel is calculated based on the following equation (5).
ver = Edge (filter_ver, OriginalL) (5)
[0092]
The filter_ver is a filter for extracting a vertical edge component, and it is preferable that the filter_ver and the filter_hor satisfy the following expression (6).
filter_ver = filter_hor ^Τ … (6)
[0093]
Here, "T" represents a transfer operation. Then, in step e2, the process returns to step b3.
[0094]
Next, the edge direction classification processing (step b4 shown in FIG. 5) will be described. FIG. 9 is a flowchart illustrating a detailed procedure of the edge direction classification processing by the edge direction classification processing unit 12. Here, based on the previously calculated horizontal edge component hor and vertical edge component ver, each pixel of the determination block is classified into any of a non-edge pixel, a horizontal edge pixel, a vertical edge pixel, or a diagonal edge pixel, The total number of edge pixels in the judgment block is calculated. This processing is performed in order from the leftmost pixel to the rightmost pixel in the upper stage in the determination block, and is repeated from each pixel in the uppermost stage to the lower right pixel in the determination block.
[0095]
First, in step f1, it is determined whether the pixel of interest is an edge pixel. This determination is based on the sum of squares (hor) of the horizontal edge component hor and the vertical edge component ver. ² + Ver ² ) Is calculated and compared with a threshold value T2. Note that the sum of squares (horror) of each edge component is used for comparison with the threshold value T2. ² + Ver ² ) Is used because the calculation result of the horizontal edge component and the vertical edge component may be a negative value, and the mere sum of the edge components (hor + ver) correctly evaluates the density change around the target pixel. It is not possible.
[0096]
(Hor ² + Ver ² Is larger than the threshold value T2, the density change around the target pixel is large, the target pixel is determined to be an edge pixel, and the process proceeds to step f3. (Hor ² + Ver ² ) Is equal to or smaller than the threshold value T2, the process proceeds to step f2. In Step f2, since the density change around the target pixel is small, it is determined that the pixel is a non-edge pixel, and the process proceeds to Step f9.
[0097]
In step f3, 1 is added to the total number of edges Cnt_total. The total number of edges Cnt_total is the total number of edge pixels in the determination block. In step f4, the squared difference value (hor) between the horizontal edge component hor and the vertical edge component ver ² -Ver ² ) Is compared with a threshold T3. (Hor ² -Ver ² ) Is larger than the threshold value T3, the process proceeds to step f5, the pixel of interest is determined to be a horizontal edge pixel, and the process proceeds to step f9. (Hor ² -Ver ² ) Is equal to or smaller than the threshold value T3, the process proceeds to step f6.
[0098]
In step f6, (hor ² -Ver ² ) Is compared with a threshold value T4 smaller than the threshold value T3. ² -Ver ² ) Is smaller than the threshold value T4, the process proceeds to step f7. In step f7, the target pixel is determined to be a vertical edge pixel, and the process proceeds to step f9. (Hor ² -Ver ² ) Is equal to or larger than the threshold value T4, the process proceeds to step f8. In step f8, the target pixel is determined to be a diagonal edge pixel, and the process proceeds to step f9. As described above, the reason why the square difference of each edge component is used for comparison with the threshold value T3 is that the calculation result of the horizontal edge component and the vertical edge component may be a negative value, This is because (hor-ver) cannot evaluate the correct edge direction.
[0099]
In step f9, it is determined whether or not the edge direction classification processing has been completed for all pixels forming the determination block. If not completed, the process returns to step f1, and if completed, the edge direction classification processing ends.
[0100]
As described above, the edge direction can be determined only by calculating the difference between the square of the horizontal edge component hor and the square of the vertical edge component ver. For example, when calculating the edge angle, the ratio of the horizontal component to the vertical component is used. Compared with the case of using the arc tangent of (1), the amount of calculation can be remarkably reduced, and the processing speed can be increased without lowering the accuracy. In addition, it is possible to suppress an increase in the scale of a circuit that performs the processing.
[0101]
This edge direction classification processing may be performed in the following procedure.
FIG. 10 is a diagram illustrating a method of classifying an edge direction by an angle. When the horizontal line direction of the image data is used as a reference, the angle between the horizontal direction and the edge is θ (0 ° ≦ θ ≦ 90 °), and θ is φ _H Horizontal edge (θ <φ _H ), Φ _V Larger than the vertical edge (θ> φ _V ), Otherwise (φ _H ≦ θ ≦ φ _V ), If it is defined as a diagonal edge, the direction of the edge can be determined by the following determination formulas (7) to (9). Specifically, when the edge angle is less than 15 degrees, the horizontal edge (θ <φ _H = 15 °), vertical edge when exceeding 75 ° (θ> φ) _V = 75 °), the edge directions can be classified with high accuracy if predetermined.
ver ² -Hor ² tan ² φ _H <0 (Horizontal; horizontal) (7)
ver ² -Hor ² tan ² φ _V > 0 (Vertical; vertical) ... (8)
otherwise (Diagonal; diagonal) ... (9)
[0102]
FIG. 11 is a flowchart illustrating another procedure of the edge direction classification process. The processing procedure is substantially the same as the processing procedure shown in FIG. 9, but the judgment criteria for judging a horizontal edge, a vertical edge, and a diagonal edge in step f4 ′ and step f6 ′ are different. That is, the above formulas (7) and (8) are used as criteria.
[0103]
Next, the direction correlation calculation processing (step b5 shown in FIG. 5) will be described. FIG. 12 is a flowchart illustrating a detailed procedure of the directional correlation calculation processing by the directional correlation calculation processing unit 13. The correlation indicates how much the direction of the edge of each pixel and the direction of the edge of the neighboring pixel are the same or different, that is, the ratio of neighboring pixels having the same edge direction among neighboring pixels. In the present embodiment, the center of a 3 × 3 pixel group is set as a target pixel, and eight surrounding pixels are set as neighboring pixels.
[0104]
When each edge pixel in the determination block is set as the target pixel, among the nearby edge pixels of the target pixel as the correlation, the number of nearby edge pixels having the same edge direction as the edge direction of the target pixel is counted for each target pixel, The number of pixels of the target pixel in which the counted number of neighboring edge pixels is smaller than a predetermined value is calculated as a feature amount. This processing is performed in order from the leftmost pixel to the rightmost pixel in the upper stage in the determination block, and is repeated from each pixel in the uppermost stage to the lower right pixel in the determination block. The processing for the pixels located at the ends of the determination block is performed by interleaving, for example, by interpolating pixel data of adjacent pixels in the determination block.
[0105]
First, in step g1, it is determined whether or not the target pixel is an edge pixel based on the determination results in steps f1 and f2 of the edge direction classification processing. If it is an edge pixel, the process proceeds to step g2, and if it is not an edge pixel, the process proceeds to step g9.
[0106]
In step g2, similarly to step g1, it is determined whether or not a predetermined neighboring pixel among the neighboring pixels of the target pixel is an edge pixel. If it is an edge pixel, the process proceeds to step g3. If it is not an edge pixel, the process proceeds to step g6. In step g3, 1 is added to the number of neighboring edges Cnt_neighbor. This makes it possible to calculate the number of neighboring pixels that are edge pixels.
[0107]
In Step g4, the edge directions are compared based on the determination results in Steps f4 to f8 in the edge direction classification processing, and it is determined whether or not the edge directions of the target pixel and the neighboring pixels are equal. If they are equal, the procedure proceeds to step g5, and if they are not equal, the procedure proceeds to step g6. At step g5, 1 is added to the number of edges Cnt_same in the same direction. As a result, the number of neighboring pixels having an edge in the same direction as the edge of the target pixel can be calculated.
[0108]
Then, in step g6, it is determined whether or not the processing has been completed for all neighboring pixels. If the processing has been completed, the process proceeds to step g7, and if not completed, the process returns to step g2 to process the next neighboring pixel.
[0109]
In Step g7, the value obtained by doubling the number Cnt_same of the same-direction edges is compared with the number Cnt_neighbor of the neighboring edges of the target pixel. When 2Cnt_same <Cnt_neighbor, the number of edges in the same direction is less than half the number of neighboring edges, so that the ratio of neighboring edge pixels having an edge direction different from the edge direction of the target pixel is large. In step g8, 1 is added to the number of correlated pixels Cnt_high_cor, which is a feature amount. Then, the values of Cnt_same and Cnt_neighbor are cleared. At step g9, it is determined whether the processing has been completed for all the pixels constituting the determination block. If the processing has been completed, the directional correlation calculation processing ends, and if not, the processing returns to step g1.
[0110]
In this way, for each pixel of interest in the pixel block, the number of neighboring edge pixels having the same edge direction as the direction of the edge of the pixel of interest is counted, and the number of pixels of interest whose count value becomes smaller than a predetermined value is calculated. Since the calculation is performed, the feature amount of the pixel block can be easily calculated.
[0111]
Note that the direction correlation calculation process can also be performed using the inner product as described below. FIG. 13 is a flowchart illustrating a processing procedure when the directional correlation calculation processing is performed using the inner product. First, the following vectors are assigned to the directions of the respective edges in advance. The target pixel is Xc, the neighboring pixels are Xn, and vector values (10) to (13) assigned to the edge direction are stored in Xc and Xn. The vector value here is a predetermined value replaced according to the direction of the edge.
Horizontal → [100] ... (10)
Vertical direction → [0 10] ... (11)
Oblique direction → [0 0 1] (12)
Non-edge → [0 0 0] (13)
[0112]
Using these vector values, the number of neighboring edges Cnt_neighbor and the number of same-direction edges Cnt_same are calculated from Expressions (14) and (15).
[0113]
[Equation 3]

[0114]
(Equation 4)

[0115]
In the specific flow, first, in step h1, the number of neighboring edges Cnt_neighbor = Cnt_neighbor + Xn · Xn is calculated, and in step h2, the number of edges in the same direction Cnt_same = Cnt_same + Xc · Xn is calculated, and this is repeated for all pixels in the pixel block. .
[0116]
In step h3, it is determined whether or not the processing has been completed for all neighboring pixels. If it has been completed, the process proceeds to step h4. If not completed, the process returns to step h1. In step h4, the value obtained by doubling the number of edges Cnt_same in the same direction is compared with the number of edges Cnt_neighbor near the target pixel. If 2Cnt_same <Cnt_neighbor, the process proceeds to step h5, and if 2Cnt_same ≧ Cnt_neighbor, the process proceeds to step h6.
[0117]
In step h5, 1 is added to the number of correlated pixels Cnt_high_cor. Then, the values of Cnt_same and Cnt_neighbor are cleared. In Step h6, it is determined whether the processing has been completed for all the pixels constituting the determination block. If the processing has been completed, the directional correlation calculation processing ends, and if not, the processing returns to step h1.
[0118]
As described above, by using the inner product for the directional correlation calculation process, it is possible to reduce the determination process during the processing procedure and to simplify the process. That is, since it is not necessary to perform the branching process in steps g1, g2, and g4 shown in FIG. 12, the directional correlation calculation process can be performed at high speed. Further, an increase in the scale of a circuit that performs the processing can be suppressed.
[0119]
Also, if the vector components of Equations (10) to (13) are assigned to each bit, and the horizontal edge is represented by “100”, the vertical edge is represented by “010”, and the oblique edge is represented by “001”, the following equation is obtained. The inner product operation of (14) and (15) can be executed by a bit operation, and the processing can be performed at a higher speed.
[0120]
Next, the determination process (step b6 shown in FIG. 5) will be described. The determination process is performed for each determination block, and the target determination block is a determination block for which the background pixel detection process, the horizontal edge component calculation process, the vertical edge component calculation process, the edge direction classification process, and the direction correlation calculation process have been completed. . By using the number of base pixels Cnt_white calculated in the base pixel detection process, the total number of edge pixels Cnt_total calculated in the edge direction classification process, and the number of highly correlated pixels Cnt_high_cor calculated in the directional correlation calculation process, the determination block becomes a base block, It is determined whether the block is a photograph block, a character block, a halftone block, or an undefined block.
[0121]
FIG. 14 is a flowchart illustrating a detailed procedure of the determination process. First, in step i1, the number of base pixels Cnt_white is compared with a threshold value T5. If the number of base pixels Cnt_white is larger than the threshold T5, the process proceeds to step i2. If the number of base pixels Cnt_white is equal to or smaller than the threshold T5, the process proceeds to step i3. In step i2, the determination block is determined to be a base block, and the determination processing ends. This is because the number of background pixels in the determination block is greater than a certain number.
[0122]
In step i3, the total number Cnt_total of edge pixels is compared with a threshold value T6. If the total number of edge pixels Cnt_total is smaller than the threshold value T6 (Cnt_total <threshold value T6), the process proceeds to step i4. If the total number of edge pixels Cnt_total is equal to or greater than the threshold value T6 (Cnt_total ≧ threshold value T6), the process proceeds to step i5. In step i4, the determination block is determined to be a photograph block, and the determination processing ends. This is because the total number of edge pixels in the determination block is smaller than a certain number, the image is a halftone image, and is not a base block.
[0123]
In step i5, the number of correlated pixels Cnt_high_cor is compared with a threshold value T7. If Cnt_high_cor is smaller than threshold T7 (Cnt_high_cor <threshold T7), the process proceeds to step i7, and if Cnt_high_cor is equal to or greater than threshold T7 (Cnt_high_cor ≧ threshold T7), the process proceeds to step i6.
[0124]
In step i7, 2Cnt_high_cor obtained by doubling the number of correlated pixels Cnt_high_cor is compared with the total number of edge pixels Cnt_total. If 2Cnt_high_cor is smaller than the total number of edge pixels Cnt_total (2Cnt_high_cor <Cnt_total), the process proceeds to step i8. In this case, the number of correlated pixels is smaller than half of the total number of edge pixels, and there are many pixels having the same edge direction in the determination block. It is determined to be a character block, and the determination processing ends.
[0125]
If 2Cnt_high_cor is equal to or larger than the total number of edge pixels Cnt_total (2Cnt_high_cor ≧ Cnt_total) in step i7, the process proceeds to step i9. In this case, contrary to the above, since there are many pixels having different edge directions in the determination block, it is possible that the block is a halftone block. However, there is a possibility that an erroneous determination is made, so to avoid this, the process proceeds to step i9, the block is determined to be an indefinite block, and the determination process ends.
[0126]
Similarly, in step i6, 2Cnt_high_cor obtained by doubling the number of correlated pixels Cnt_high_cor is compared with the total number of edge pixels Cnt_total. If 2Cnt_high_cor is smaller than the total number of edge pixels Cnt_total (2Cnt_high_cor <Cnt_total), a large number of pixels having the same direction of the edge exist in the determination block, and thus it may be a character block. However, there is a possibility that an erroneous determination is made, so to avoid this, the process proceeds to step i9, the block is determined to be an indefinite block, and the determination process ends.
[0127]
If 2Cnt_high_cor is equal to or greater than the total number of edge pixels Cnt_total (2Cnt_high_cor ≧ Cnt_total), the process proceeds to step i10. In this case, since there are many edges in different directions in the determination block, in step i10, the determination block is determined to be a halftone dot block, and the determination process ends.
[0128]
Through the above-described determination processing, the determination block can be separated into a character block, a halftone block, a photograph block, a background block, and an undefined block.
[0129]
By performing each of the above-described processes for all determination blocks in the image data, it is possible to perform a region separation process on the image data.
[0130]
As described above, by using the feature amount based on the direction of the edge of the pixel, it is possible to perform the region separation based on the structural feature of the edge direction of the image. As a result, it is possible to accurately separate even areas having very similar frequency distributions.
[0131]
In the above, the method of performing region separation on a pixel block having a predetermined number of pixels has been described. However, the region may be determined on a target pixel in a pixel block.
[0132]
FIG. 15 is a block diagram illustrating a configuration of the image forming apparatus Y. The image forming apparatus Y includes an image processing apparatus X, an image input device 60, and an image output device 61, and is used as, for example, a digital copying machine or a multifunction machine using an electrophotographic system or an inkjet system.
[0133]
The above-described area separation processing unit A is provided in the image processing apparatus X. As a result, the image processing apparatus X performs a region separation process on the input image data and identifies the region of each block constituting the image data, thereby performing various image processes according to the characteristics of the region, for example, a filter process. , Gamma processing, and halftone processing. Accordingly, a high-quality image can be obtained, and the image forming apparatus Y can form a high-quality image from the image data on which the image processing has been performed by the image processing apparatus X.
[0134]
The image input device 60 reads an input image and converts it into image data. The image input device 60 includes a scanner unit having a CCD (Charge Coupled Device) and the like, and reads a reflected light image from the image by the CCD. Is converted into an RGB analog signal for each pixel, and this analog signal is input to the image processing apparatus X.
[0135]
The image output device 61 forms an image on a recording medium such as paper based on the image data processed by the image processing device X. The image formation is performed by, for example, an electrophotographic system or an inkjet system.
[0136]
Next, the image processing apparatus X will be described. The image processing apparatus X includes an analog / digital (hereinafter abbreviated as “A / D”) converter 50, a shading correction unit 51, an input gradation correction unit 52, a region separation processing unit A, a color correction unit 53, and a black generation unit. It comprises a color removal processing section 54, a spatial filter processing section 55, and a halftone output gradation processing section 56.
[0137]
The image data of the RGB analog signal input from the image input device 60 is converted into an A / D converter 50, a shading correction unit 51, an input gradation correction unit 52, an area separation processing unit A, a color correction unit 53, and a black generation unit. The color removal processing unit 54, the spatial filter processing unit 55, and the halftone output gradation processing unit 56 are processed in this order, and as a CMYK (C: cyan, M: magenta, Y: yellow, K: black) digital color signal, The image is output to the image output device 62.
[0138]
The A / D converter 50 converts an RGB analog signal input from the image input device 60 into a digital signal, and outputs the digital signal to the shading corrector 51. The shading correction unit 51 performs a process of removing various types of distortion generated in the illumination system, the imaging system, and the imaging system of the image input device 60 on the RGB digital signal input from the A / D conversion unit 50, and performs RGB processing. Is output as a reflectance signal.
[0139]
The input tone correction unit 52 adjusts the color balance of the RGB reflectance signals from which various distortions have been removed by the shading correction unit 51, and also controls the image processing system employed in the image processing apparatus X such as density. To a density signal suitable for
[0140]
Based on the RGB signal image data input from the input tone correction unit 52, the area separation processing unit A converts the image data for each block or pixel for each character, halftone dot, photograph, background, Area separation processing is performed to separate the areas. The area separation processing unit A outputs an area identification signal indicating which area each block or each pixel belongs to to the black generation under color removal processing unit 54 and the spatial filter processing unit 55 based on the processing result. The RGB digital signal input from the input tone correction section 52 is output to the color correction section 53 as it is. The region identification signal may be output to the halftone output gradation processing unit 56.
[0141]
The color correction unit 53 converts the RGB density signals into CMY (C: cyan, M: magenta, Y: yellow) density signals, and performs color correction processing on the CMY density signals in order to faithfully reproduce colors. Is applied. Specifically, the color correction process is a process of removing color turbidity based on the spectral characteristics of CMY toner and ink containing unnecessary absorption components from the CMY density signal.
[0142]
The black generation and under color removal processing unit 54 uses the area identification signal output from the area separation processing unit A, and outputs a black (K) black based on the CMY color signals output from the color correction unit 53 for each area. A black generation process for generating a color signal and an under color removal process for removing an under color from the CMY color signal are performed. The under color removal process is a process of obtaining a new CMY color signal by subtracting the black color signal generated by the black generation process from the CMY color signal. As a result of these processes, the CMY density signals are converted into image data composed of CMYK color signals.
[0143]
The spatial filter processing unit 55 performs a spatial filter process using a digital filter on the CMYK image data obtained by the black generation and under color removal processing unit 54. This processing is also performed for each area using the area identification signal output from the area separation processing unit A. As a result, the spatial frequency characteristics of the image are corrected, so that it is possible to prevent the image output from the image output device from being blurred or having graininess deterioration.
[0144]
The halftone output gradation processing section 56 performs a halftone generation process. The halftone generation process is a process in which an image is divided into a plurality of image tones so that a tone can be reproduced, and a binary or multivalued dither method, an error diffusion method, or the like can be used. Note that the halftone output gradation processing unit 56 may perform a process of converting the density value of the image data into a dot area ratio which is a characteristic value of the image output device 61. Then, the density signal processed by the halftone output gradation processing section 56 is given to the image output device. The operation of the image processing apparatus X is controlled by, for example, a CPU (Central Processing Unit) included in the image processing apparatus X.
[0145]
The present invention is realized by a program for executing the above-described area separation processing method. This program may be stored directly in the main memory of the device that executes the image processing method, or may be stored in a recording medium that can be read by inserting it into a fixed disk device. This makes it possible to provide a portable recording medium on which a program for executing the area separation processing method is recorded.
[0146]
In the present embodiment, as the recording medium, a memory, for example, a ROM (Read Only Memory) itself may be a program medium in order to be processed by the microcomputer, Alternatively, a program reading device may be provided as an external storage device, and may be a readable program medium that can be inserted therein.
[0147]
In any case, the stored program may be configured to be accessed and executed by a microprocessor, or in any case, the program may be read and the read program may be stored in a program storage of the microcomputer. The program may be downloaded to an area and the program may be executed. It is assumed that the download program is stored in the main unit in advance.
[0148]
Here, the program medium is a recording medium configured to be separable from the main body, such as a tape system such as a magnetic tape or a cassette tape, a magnetic disk such as a flexible disk or a hard disk, a CD-ROM (Compact Disk Read Only Memory), Disk systems of optical disks such as MO (Magneto Optical) disks, MDs (Mini Disks) and DVDs (Digital Versatile Disks), card systems such as IC (Integrated Circuits) cards including memory cards and optical cards, mask ROMs, EPROMs (Erasables) Programmable Read-Only Memory, EEPROM (Electrically Erasable Program) mmable Read-Only Memory) and a flash ROM may be a medium fixedly carrying the program, including semiconductor memories due.
[0149]
Further, in the case where the present embodiment has a system configuration capable of connecting to a communication network including the Internet, the medium may carry a program in a fluid manner so as to download the program from the communication network. When the program is downloaded from the communication network as described above, the download program may be stored in the main device in advance, or may be installed from another recording medium.
[0150]
The above-described recording medium is read by a program reading device provided in a digital image forming apparatus or a computer system, so that the above-described area separation processing method is executed.
[0151]
The computer system includes an image input device 60 such as a flatbed scanner, a film scanner, and a digital camera, a computer on which various processing such as an image processing method is performed by loading a predetermined program, and a CRT display for displaying a processing result of the computer. -It is composed of an image display device such as a liquid crystal display and a printer that outputs the processing results of the computer to paper or the like. Further, a modem or the like is provided as communication means for connecting to a server or the like via a network.
[0152]
【The invention's effect】
As described above, according to the present invention, segmentation is performed based on the structural feature of an image, which is the direction of an edge. Therefore, it is possible to accurately separate a halftone dot region and a character region in which spatial frequency components are approximated. Can be. In addition, since the background area is determined using the detected number of background pixels, the background area can be easily separated. In addition, since the determination of the photograph area is performed using the detected number of edge pixels, the photograph area can be easily separated.
[0153]
Further, according to the present invention, the direction of the edge can be determined only by calculating the difference between the horizontal edge component and the vertical edge component, so that the calculation amount can be reduced and the speed of the region separation processing can be increased without lowering the accuracy. Can be achieved. Further, the circuit scale for performing the region separation processing can be reduced.
[0154]
Further, according to the present invention, since the feature amount is calculated based on the number of neighboring edge pixels having the same edge direction as the edge direction of the target pixel, the feature amount of the pixel block can be easily calculated.
[0155]
According to the present invention, the edge direction of the edge pixel is replaced with a predetermined value according to the direction, and the number of neighboring edge pixels having the same edge direction as the edge direction of the target pixel is calculated. It is not necessary to perform the calculation, the amount of calculation can be reduced, and the speed of the region separation processing can be increased. Further, the circuit scale for performing the region separation processing can be reduced.
[0156]
Further, according to the present invention, an area separation processing program for causing a computer to execute the above-described area separation processing method can be provided.
[0157]
Further, according to the present invention, it is possible to provide a computer-readable recording medium that records an area separation processing program for causing a computer to execute the above-described area separation processing method.
[0158]
Further, according to the present invention, it is possible to perform high-accuracy region separation and output image data that has been subjected to processing according to the region, so that a high-quality image can be formed.
[Brief description of the drawings]
FIG. 1 is a diagram showing a distribution of an edge direction in a character image and a halftone image.
FIG. 2 is a block diagram illustrating a configuration of a region separation processing unit A that performs a region separation process.
FIG. 3 is a flowchart illustrating a procedure of a region separation process performed by a region separation processing unit A;
FIG. 4 is a block diagram showing a configuration of a block determination unit 3 shown in FIG.
FIG. 5 is a flowchart illustrating a procedure of a block determination process performed by a block determination unit 3;
FIG. 6 is a flowchart showing a detailed procedure of a background pixel detection process by a background pixel detection processing unit 10;
FIG. 7 is a flowchart illustrating a procedure of a horizontal edge component calculation process by a horizontal edge component calculation processing unit 111;
FIG. 8 is a flowchart illustrating a procedure of a vertical edge component calculation process by a vertical edge component calculation processing unit 112;
FIG. 9 is a flowchart illustrating a detailed procedure of an edge direction classification process by an edge direction classification processing unit;
FIG. 10 is a diagram illustrating a method of classifying an edge direction by an angle.
FIG. 11 is a flowchart illustrating another procedure of the edge direction classification process.
FIG. 12 is a flowchart illustrating a detailed procedure of a direction correlation calculation process by a direction correlation calculation processing unit 13;
FIG. 13 is a flowchart illustrating a processing procedure when a directional correlation calculation process is performed using an inner product.
FIG. 14 is a flowchart illustrating a detailed procedure of a determination process.
FIG. 15 is a block diagram illustrating a configuration of an image forming apparatus Y.
[Explanation of symbols]
1 Block division
2 L ^* a ^* b ^* Conversion unit
3 Block judgment section
10 Background pixel detection processing unit
11 Edge component calculation processing unit
12 Edge direction classification processing unit
13 direction correlation calculation processing unit
14 Judgment processing unit
50 analog / digital (A / D) converter
51 Shading correction unit
52 Input tone correction unit
53 color correction unit
54 Black Generation Under Color Removal Processing Unit
55 spatial filter processing unit
56 Halftone output gradation processing section
60 Image input device
61 Image output device
A area separation processing unit
X image processing device
Y Image forming device

Claims (10)

Image data representing an image composed of a plurality of pixels is input, and based on the input image data, a pixel block composed of a plurality of pixels or a pixel of interest in the pixel block is divided into a plurality of regions including a character region and a halftone dot region. In the area separation processing method of determining which one belongs to and performing area separation of image data,
A dividing step of dividing the image data into pixel blocks each having a predetermined number of pixels;
A component calculation step of calculating an edge component of each pixel in the pixel block,
A direction determining step of determining a direction of an edge formed by each pixel based on the calculated edge component;
A feature amount calculating step of calculating a feature amount of the pixel block based on the correlation between the direction of the edge of each pixel in the pixel block and the direction of the edge of a neighboring pixel based on the determined direction of the edge;
Determining whether the pixel block or the pixel of interest in the pixel block belongs to the character area or the halftone dot area based on the calculated feature amount. . Image data representing an image composed of a plurality of pixels is input, and based on the input image data, a plurality of pixel blocks including a plurality of pixels or a plurality of pixels of interest in a pixel block including a character area, a halftone area, and a photograph area. In the region separation processing method of determining which of the regions belongs to, and performing region separation of the image data,
A dividing step of dividing the image data into pixel blocks each having a predetermined number of pixels;
A component calculation step of calculating an edge component of each pixel in the pixel block,
An edge pixel counting step of detecting an edge pixel in the pixel block based on the edge component and counting the number of edge pixels;
A direction determining step of determining a direction of an edge formed by each edge pixel based on the edge component,
A feature value calculating step of calculating a feature value of the pixel block based on the correlation between the direction of the edge of each edge pixel in the pixel block and the direction of the edge of its neighboring edge pixels based on the determined direction of the edge;
Based on the counted number of edge pixels, it is determined whether or not the pixel of interest in the pixel block or the pixel block belongs to a photographic area, and based on the calculated feature amount, A determining step of determining whether or not a pixel belongs to the character area or the halftone dot area. Image data indicating an image composed of a plurality of pixels is input, and based on the input image data, a pixel block composed of a plurality of pixels or a pixel of interest in the pixel block is a character area, a halftone dot area, a photograph area, and a base area. In a region separation processing method of determining which of a plurality of regions including a region belongs, and performing region separation of image data,
A dividing step of dividing the image data into pixel blocks each having a predetermined number of pixels;
A base pixel counting step of detecting a base pixel in the pixel block based on the tone value of the pixel and counting the number of base pixels;
A component calculation step of calculating an edge component of each pixel in the pixel block,
An edge pixel counting step of detecting an edge pixel in the pixel block based on the edge component and counting the number of edge pixels;
A direction determining step of determining a direction of an edge formed by each edge pixel based on the edge component,
A feature value calculating step of calculating a feature value of the pixel block based on the correlation between the direction of the edge of each edge pixel in the pixel block and the direction of the edge of its neighboring edge pixels based on the determined direction of the edge;
It is determined whether or not the pixel of interest in the pixel block or the pixel block belongs to the background area based on the counted number of background pixels, and the pixel of interest in the pixel block or pixel block is determined based on the counted number of edge pixels. Determining whether the pixel block belongs to a photograph area, and determining whether the pixel of interest in the pixel block or the pixel block belongs to the character area or the halftone dot area based on the calculated feature amount; and A region separation processing method comprising: The component calculation step calculates a horizontal edge component and a vertical edge component as edge components,
4. The method according to claim 1, wherein the direction determining step determines an edge direction from a difference between the horizontal edge component and the vertical edge component. 5. In the feature amount calculating step, when each edge pixel in the pixel block is set as a target pixel, among the neighboring edge pixels of the target pixel, the number of neighboring edge pixels having the same edge direction as the edge direction of the target pixel is determined for each target pixel. Counting,
The region separation processing method according to any one of claims 1 to 4, wherein the number of pixels of the target pixel in which the counted number of neighboring edge pixels is smaller than a predetermined value is calculated as a feature amount. In the feature amount calculation step, the direction of the edge of the edge pixel is replaced with a predetermined value according to the direction,
6. The method according to claim 5, wherein the number of neighboring edge pixels having the same edge direction as the target pixel is counted using the replaced value. An area separation processing program for causing a computer to execute the area separation processing method according to any one of claims 1 to 6. A computer-readable recording medium storing an area separation processing program for causing a computer to execute the area separation processing method according to claim 1. Image data representing an image composed of a plurality of pixels is input, and based on the input image data, a pixel block composed of a plurality of pixels or a pixel of interest in the pixel block is divided into a plurality of regions including a character region and a halftone dot region. In an image processing apparatus including an area separation processing unit that determines which one belongs to, and performs area separation of image data,
Dividing means for dividing the image data into pixel blocks each having a predetermined number of pixels;
Component calculating means for calculating an edge component of each pixel in the pixel block,
Direction determining means for determining the direction of the edge formed by each pixel based on the calculated edge component,
A feature amount calculating unit configured to calculate a feature amount of the pixel block based on a correlation between the direction of the edge of each pixel in the pixel block and the direction of the edge of a neighboring pixel based on the determined direction of the edge;
An image processing apparatus comprising: a determination unit configured to determine whether the pixel block or the pixel of interest in the pixel block belongs to the character region or the halftone dot region based on the calculated feature amount. An image processing apparatus according to claim 9,
An image forming apparatus comprising: an image output device that outputs image data processed by the image processing device.

Images