KR100683060B1 - Apparatus and method for removing blocking of image frames - Google Patents

Abstract

The present invention relates to an apparatus and method for removing blocking of a decoded image frame.

According to an embodiment of the present invention, there is provided a method of eliminating blocking of an image frame, by applying an edge detector to an arbitrary pixel in an image frame decoded in units of blocks, extracting pixel information including an edge response size, edge direction information, and edge position information. And classifying the pixel into one of non-edge pixels, object edge pixels, and block edge pixels using the extracted pixel information, and performing different filtering according to the classification result.

According to the configuration of the present invention, it is possible to provide an image of improved quality by selectively preserving the object edge information.

Block, Edge, Filter

Description

Blocking phenomenon of video frame and method thereof {DEVICE AND METHOD FOR DEBLOCKING OF VIDEO FRAME}

1 is a block diagram illustrating a video encoding apparatus according to an embodiment of the present invention.

2 is a block diagram illustrating a deblocking filter according to an embodiment of the present invention.

3 is a flowchart illustrating an edge characteristic determination method according to an exemplary embodiment of the present invention.

FIG. 4 is a diagram schematically illustrating a blocking phenomenon occurring at a block boundary in a homogenous region.

5 is a diagram illustrating block edges and object edges in decoded image data.

6 is a flowchart illustrating a selective filtering method of image data according to an embodiment of the present invention.

The present invention relates to an apparatus and method for removing blocking of a decoded image frame.

Most conventional image data encoding methods process image frames in units of blocks. In particular, according to encoding standards for image data such as MPEG (H.264) and H.264, encoding is performed in units of blocks. That is, in the encoding process of the image frame, the frame is divided into blocks and individual data processing is performed on each of the divided blocks, thereby causing a blocking phenomenon in which blocking artifacts are generated due to the existence of outlines of block boundaries. Therefore, a problem arises that the image quality deterioration and the user's visual inconvenience.

This blocking phenomenon is caused by the loss of low frequency component information due to encoding and quantization processes that do not take into account the correlation between the divided blocks. Refers to a phenomenon in which a high frequency region, that is, an edge region, does not exist. Since the blocking phenomenon destroys the continuity between adjacent pixels, a deblocking process for smoothing a high frequency region using a lowpass filter to remove the blocking phenomenon has been generally performed.

In addition to the deblocking process using low-band filtering, a method of classifying pixels in one block into two types of pixels inside a block and block boundary pixels, and accordingly, a method of varying the strength of low-band filtering has been proposed. In addition, among the pixels corresponding to the block boundary, a set of pixels whose boundary value is larger than a certain threshold value is selected, and a virtual image frame set so that the values can be canceled with each other is appropriately inserted to smoothly change all pixel values. In addition, other forms of low-band filtering have been proposed that reduce visual discontinuities.

However, the above-described conventional blocking effect removal techniques can reduce visual discontinuity appearing at the boundary of the block, but the high frequency component generated by the contour of the person or object located around the boundary area of the block, that is, the object edge Edge information is also smoothed together, resulting in a deterioration in image quality of the entire image.

That is, in order to perform filtering to remove the blocking phenomenon, it is important to first identify the region where the blocking phenomenon occurs. Otherwise, the image quality after the filtering may be deteriorated by determining that the edge area existing in the original image frame, that is, the object edge, is removed by the blocking phenomenon.

An object of the present invention is to provide an apparatus and method for selectively filtering an image frame capable of preserving object edges in an image and effectively removing a blocking phenomenon in reconstructing an image frame encoded on a block basis.

In order to solve this problem, according to an aspect of the present invention, an edge detector is applied to an arbitrary pixel in an image frame decoded on a block basis to obtain pixel information including the magnitude of edge response, edge direction information, and edge position information. Extracting, classifying the pixel into one of non-edge pixels, object edge pixels, and block edge pixels using the extracted pixel information, and performing different filtering according to the classification result. A method of removing blocking phenomenon of is provided.

Here, the pixel classification step compares the magnitude of the edge response of the pixel and a preset threshold value, and classifies the pixel as an unedge pixel when the threshold value is large. When the magnitude of the edge response is large, it is determined whether the edge direction of the pixel corresponds to 0 ° or 90 °. When the edge direction is not 0 ° or 90 °, the pixel is an object edge pixel. When the edge direction corresponds to 0 ° or 90 °, it is determined whether the pixel exists in the block boundary area by using edge position information of the pixel, and the pixel is located in the block boundary area. If so, it is classified as a block edge pixel.

According to another feature of the invention, the pixel information extraction unit for extracting the pixel information including the magnitude of the edge response, edge direction information and edge position information by applying the edge detector to any pixel in the image frame decoded in block units; Blocking an image frame including an edge discriminator classifying the pixel into one of non-edge pixels, object edge pixels, and block edge pixels using the extracted pixel information, and a filter unit performing different filtering according to the pixel classification result. A development removal device is provided.

According to still another aspect of the present invention, there is provided a computer-readable recording medium having recorded thereon a program for realizing the above-described method.

DETAILED DESCRIPTION Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings so that those skilled in the art may easily implement the present invention. As those skilled in the art would realize, the described embodiments may be modified in various different ways, all without departing from the spirit or scope of the present invention.

In the drawings, parts irrelevant to the description are omitted in order to clearly describe the present invention, and like reference numerals designate like parts throughout the specification.

1 is a block diagram illustrating a video encoding apparatus according to an embodiment of the present invention. As shown in FIG. 1, a video encoding apparatus according to an exemplary embodiment of the present invention includes a transformer 110, a quantizer 120, an inverse quantizer 131, an inverse transformer 132, a frame memory 133, and a motion compensator. 134, an adder 135, a motion estimator 140, a subtractor 150, an entropy encoder 160, and a deblocking filter 200. In this case, the video encoding apparatus of FIG. 1 may be an H.264 encoding apparatus.

The image frame input in macroblock units is converted according to a preset method in the converter 110 and then quantized in the quantizer 120. In this case, the image data transformation technique used in the transformer 110 may be a discrete cosine transform (DCT) or a discrete wavelet transform (DWT) technique.

The quantized image frame output from the quantizer 120 is decoded through the inverse quantizer 131 and the inverse transformer 132, and the decoded data is input to the adder 135.

The decoded data output from the adder is restored to an actual image through the pixel information extraction, the edge characteristic determination, and the selective filtering according to the determination result by the deblocking filter 200 and stored in the frame memory 133. Here, the reconstructed image frame stored in the frame memory 133 may be provided as a reference image for motion estimation. The filtering based on the edge determination result of the deblocking filter 200 will be described in detail below.

The motion estimator 140 performs motion estimation by receiving at least one of the reference images stored in the frame memory 133 with respect to the image frame in the macroblock unit that is currently input, and performs a motion vector and a reference image. The motion data including the index and the block mode are output.

The motion compensator 134 extracts and outputs an image frame in macroblock units corresponding to a macroblock currently input from a reference image stored in the frame memory 133 according to the motion data input from the motion estimator 140. The data output from the motion compensator 134 is input to the adder 135, reconstructs the decoded data input from the other end of the adder 135 to an actual image, and transmits the decoded data to the deblocking filter 200 to block blocking defects. Allows the removal process to be made.

When the interframe predictive coding is performed on the macroblock input to the video encoding apparatus, the subtractor 150 receives a macroblock in the reference image corresponding to the input macroblock from the motion compensator 134. The residual signal is output by performing a difference operation with the block. The output residual signal is transformed and quantized through the transformer 110 and the quantizer 120, entropy-encoded by the entropy encoder 160, and output in the form of network abstraction layer (NAL) unit data.

The image frame data output from the entropy encoder 160 may be temporarily stored in a buffer (not shown) and then output in order to adjust a bit rate.

In the video encoding apparatus illustrated in FIG. 1, an inverse quantizer 131, an inverse transformer 132, a frame memory 133, a motion compensator 134, an adder 135, and a deblocking filter 200 are included in a video encoding apparatus. A decoder 130 for decoding the encoded image data therein is formed.

2 is a block diagram illustrating a deblocking filter 200 according to an exemplary embodiment of the present invention, and FIG. 3 is a flowchart illustrating an edge characteristic determination process performed by the edge determination unit 220 of FIG. 2. As shown in FIG. 2, the deblocking filter 200 includes a pixel information extracting unit 210, an edge discriminating unit 220, and a filter unit 230, and the filter unit 230 includes a low band filter 231. ) And adaptive filter 232. According to an embodiment of the present invention, in the conventional block elimination techniques, edge information of objects located around a boundary of a block existing in an original image is low-band filtered together with edge information of a block boundary region. In order to prevent the deterioration of edge information due to deterioration of the edge information, the decoded raw image data is analyzed to extract edge characteristic information included in each pixel, and selective filtering is performed based on this.

The pixel information extracting unit 210 extracts edge information included in the pixel using an edge detector as a previous step for determining an edge pixel in the raw image data that has undergone inverse transformation and inverse quantization. The edge detector can be any of the commonly used edge detectors, including Sobel edge detectors and Canny edge detectors. The pixel information extracting unit 210 extracts the following three pieces of pixel information to effectively grasp the edge information used to determine the pixel characteristics.

1. Magnitude of edge response (M)

It can be defined as the absolute value of the response size applying the edge detector to the decoded raw image data, it can be obtained through the following equation (1).

In Equation 1, (i, j) denotes a position of an arbitrary pixel in the data of a block unit to be decoded and input, and M (i, j) represents the magnitude of an edge response of the pixel. f (i, j) denotes a pixel value representing brightness, luminance, color, etc. of an arbitrary pixel in the inputted raw image, and a value represented by f (i, j) is a device to which a video encoding apparatus is applied. It may be appropriately selected and used according to the characteristics of the and its operation policy (operation policy).

K (m, n) represents a two-dimensional edge detection operator, and (m, n) is a derivative operator as an index inside an edge mask used for edge detection. Edge is a part of differential filter, because the edge is the part of the image having a sharp change in color and a sudden change in color, and thus a derivative operation taking a change in function can be used to extract the contour. In other words, the edge of a pixel having a large rate of change of brightness value, that is, a differential value along the horizontal axis and the vertical axis, is the edge. In order to take advantage of this property, edge detection is achieved by applying a 3x3 mask consisting of a differential operator with respect to a particular pixel in the image data. For example, the Sobel edge operator can be used, and the Sobel edge operator decodes two types of edge masks: an edge mask for horizontal edge detection and an edge mask for vertical edge detection, each consisting of nine differential operators. Applied to the generated video data. The derivative operator components in the Sobel edge mask are as shown in Tables 1 and 2 below, and are the derivative operators in which the components of the second row and the second column match the target pixels for detecting edge components in the horizontal and vertical directions, respectively.

Horizontal edge detector -One 0 One -2 0 2 -One 0 One

Vertical edge detector -One -2 -One 0 0 0 One 2 One

2. Edge direction (θ)

와

를 이용하는 아래의 수학식 2를 이용하여 정의한다.It means the ratio of vertical component and horizontal component included in the pixel, and each response component with respect to the horizontal and vertical axis of the edge detector obtained by applying the edge detector with respect to one specific pixel

Wow

It is defined using Equation 2 below.

는 수평 에지 검출자를 영상 데이터에 적용한 값이다. 특정한 하나의 화소를 중심으로 하는 3×3 화소 블록에 표 1에 나타낸 바와 같은 수평 방향의 소벨 에지 검출자를 이용하는 경우를 예를 들어 설명한다. 화소 정보 추출부(210)로 입력된 원시 디코딩 데이터 내의 3×3 블록의 화소 성분을 제1행부터 아래 방향으로 각각 [a,b,c], [d,e,f], [g,h,k]라 한다면, 화소 정보를 추출하고자 하는 대상 화소인 (i,j) 화소는 블록 중심에 위치한 'e' 값을 갖는 화소가 되며, 이 때의

는 {((-1)×a)+(0×b)+(1×c)+((-2)×d)+(0×e)+(2×f)+((-1)×g)+(0×h)+(1×k)}의 값을 갖게 된다.

성분도 표 2에 나타낸 수직 방향의 소벨 에지 검출자를 이용하여 같은 방법으로 구할 수 있다.here,

Is a value obtained by applying a horizontal edge detector to image data. An example will be described using a Sobel edge detector in the horizontal direction as shown in Table 1 in a 3x3 pixel block centered on one particular pixel. The pixel components of the 3x3 block in the raw decoded data inputted to the pixel information extracting unit 210 are [a, b, c], [d, e, f], [g, h] from the first row to the downward direction, respectively. , k], the (i, j) pixel, which is the target pixel from which the pixel information is to be extracted, becomes a pixel having an 'e' value located at the center of the block.

Is (((-1) × a) + (0 × b) + (1 × c) + ((-2) × d) + (0 × e) + (2 × f) + ((-1) × g) + (0 × h) + (1 × k)}.

The components can also be obtained in the same manner using the Sobel edge detector in the vertical direction shown in Table 2.

,

와 수학식 2를 이용하여 얻은 θ(i,j) 성분이 0° 또는 90°에 해당하는 경우에는 해당 화소가 수평축 방향 또는 수직축 방향의 성분을 포함하고 있으므로, 블록의 경계 영역 또는 객체의 모서리 부분에 해당하는 것으로 판단할 수 있다.Obtained in this way

,

When the θ (i, j) component obtained by using Equation 2 and Equation 2 corresponds to 0 ° or 90 °, the pixel includes components in the horizontal or vertical axis direction, and thus the boundary region of the block or the edge of the object. It can be judged as

3. Edge position ([i, j])

[I, j], which is the edge position information, is defined as coordinate values for the vertical axis and the horizontal axis of the image data in the unit of the block decoded and input to the pixel information extracting unit 210. The pixel boundary is defined using the position information. It can be determined whether or not it corresponds to an area.

The edge discrimination unit 220 extracts the three kinds of pixel information extracted from the pixel information extracting unit 210, that is, the edge response size (M), the edge direction (θ), and the edge position information ([i, j]). It is determined whether or not the pixel corresponds to the edge region by using. Since the image data is a set of pixels with edge information and pixels without edge information, pixels of a uniform area having no edge information may exist in the process of extracting edge information. That is, the edge determination unit 220 encodes the input image data in a non-edge pixel corresponding to a uniform area, an object edge pixel corresponding to an outline of a person or an object, and a block unit. In order to preserve the edge information of the object, the pixel may be classified into three types of block edge pixels corresponding to the divided boundary region, and pixels including the edge information that do not correspond to the pixel may be generated. Allow selective filtering to be done.

Hereinafter, the pixel characteristic determination method will be described in detail with reference to the edge determination process illustrated in FIG. 3.

To determine the edge characteristics, the pixel information extracting unit 210 obtains three kinds of pixel information obtained by using the edge detector, that is, the magnitude M of the edge response, the edge direction θ, and the edge position [i, j]. ) Information is input to the edge determination unit 220, the edge determination unit 220 starts to determine the edge characteristic for any pixel (S310).

First, the edge discriminator 220 may determine that the edge response magnitude M of the selected pixel corresponds to a pixel in a relatively uniform area where no contour of a person or a thing exists, and is equal to or less than a preset threshold. In this case, it is determined that the pixel is a non-edge pixel, and the pixel data is transmitted to the filter unit 230 without additional processing (S320 and S360). 4 shows a blocking phenomenon occurring in a homogeneous region. In FIG. 4, one grating represents one pixel and is set to be decoded in units of 4 × 4 blocks. In the general deblocking process, the target area is an area corresponding to 2 pixels each in both directions about the 'A' line which is the boundary area of the unit block, so the area marked with 'X' is the target for vertical edge processing. This corresponds to the deblocking target pixel region. As shown in FIG. 4, since the magnitude of the edge response changes rapidly at the block boundary, the edge determination unit 220 sets a threshold value according to the characteristics of the image frame input to the video encoding apparatus, thereby providing block edges. It is used to determine the pixel.

On the other hand, if the magnitude (M) of the edge response of the selected pixel is larger than the preset threshold, the corresponding pixel corresponds to the object edge pixel or the block edge pixel. Therefore, it is determined whether the edge direction θ corresponds to the horizontal axis direction or the vertical axis direction. It is determined whether or not (S330). As shown in FIG. 4, since the block edge pixel positioned in the boundary region of each unit block has an edge component only in the horizontal axis direction or the vertical axis direction, it is determined whether θ is 0 ° or 90 °.

If it is determined that the selected pixel has an edge component in the axial direction, it is most likely to be a block edge pixel. However, the edge discriminator 220 determines the edge position information ([i j]) to determine whether the pixel corresponds to the block boundary area, and if the pixel corresponds to the block boundary area, the pixel is determined to be a block edge pixel, and the filter unit 230 does not need to process data. Output to (S340, S360). This means that even if θ corresponds to 0 ° or 90 °, there may be a pixel corresponding to the outline of the object, that is, the object edge, so that the pixels are determined to be the block edges and filtered through the low pass filter to prevent the image quality from deteriorating. It is to.

If the selected pixel does not have an edge component in the axial direction (S330) and is not located in the block boundary area even though it has an edge component in the axial direction (S340), the edge discriminator 220 determines that the pixel is an object edge pixel. In operation S350 and S360, the process is determined to determine whether the object is an edge of the object.

5 shows examples of such block edge pixels and object edge pixels. As can be seen from the contents shown in FIG. 4, in the general deblocking process, in the two unit blocks based on the line 'B', all regions marked with 'X', 'O', and 'Δ' are blocked. It corresponds to the deblocking target pixel area for removing the phenomenon. However, in the exemplary embodiment of the present invention, an area marked with 'O' is classified as an outline region of an object having an edge component in the horizontal or vertical axis direction, that is, an object edge pixel, and thus is not classified as a deblocking target pixel. The areas marked with '△' correspond to non-edge pixels as pixels corresponding to relatively uniform areas inside the object.

In the case of determining the characteristics of the pixel by using only the magnitude of the edge response and the component of the edge direction, an object edge pixel corresponding to the region shown in FIG. 5 may be classified as a block edge pixel and thus undergo a low band filtering process. In the present invention, the pixel characteristic is determined by additionally using edge position information.

In the H.264 method, since encoding and decoding are performed in units of 4x4 blocks, it is determined whether edge position values are located in block boundary regions such as [1,1], [1,2], etc. in one unit block. Check to determine the block edge pixel. In the case of using another coding scheme, a block unit that is a reference for decoding may be different, and thus, an edge position value, which is a reference for determining that the block boundary region corresponds, may also be changed.

The filter unit 230 includes a low band filter 231 and an adaptive filter 232. The low pass filter 231 removes blocking artifacts by reducing blocking artifacts for the block edge pixels. The low-pass filter 231 receives a pixel determined to be a non-edge pixel or a block edge pixel by the edge discriminator 220, and performs low-band filtering on the low-pass filter 231. Since the non-edge pixel input to the low band filter 231 includes almost no high frequency region, the pixel information is not damaged even though low band filtering is performed. As the low band filter 231, an in-loop deblocking filter defined in the H.264 standard may be used.

The adaptive filter 232 performs selective filtering on pixels determined to be object edge pixels as a result of the classification by the edge discriminator 220. The adaptive filter 232 is to preserve the information of the object edge pixel, and selectively filters only the pixels to which the identification code indicating the object edge pixel is added among the decoded data input from the edge discrimination unit 220. Since the adaptive filter 232 is intended to prevent the loss of object edge information, an all-pass filter with a gain of 1 can be used, and a sharpening filter used to enhance the edge component can be used. It can also be used.

In the exemplary embodiment of the present invention, the non-edge pixel is input and processed by the low band filter 231, but may be input and processed by the adaptive filter 232.

6 is a flowchart illustrating a selective filtering method of image data according to an embodiment of the present invention.

The image frame compressed in units of blocks is converted into raw decoded data through inverse quantization and inverse transformation (S610).

The pixel information extracting unit 210 receives the raw decoded data and receives three kinds of pixel information, namely, an edge response size (M), an edge direction (θ), and an edge position ([i , j]) information is extracted (S620).

The edge determining unit 220 is included in the raw decoded data by using response size, edge direction, and edge position information sequentially according to the process shown in FIG. 3 based on the pixel information obtained by the pixel information extracting unit 210. The pixels are classified into non-edge pixels, block edge pixels, and object edge pixels, respectively, and an additional identification code is added to the object edge pixels (S630).

The filter unit 230 determines whether an identification code is added to the pixel input from the edge determination unit 220, and if the corresponding pixel is an object edge pixel, filters the adaptive filter 232 using the adaptive filter 232 (S640, FIG. S642). On the other hand, when the pixel is not the object edge pixel, the deblocking process of removing the high frequency region component is performed through the low band filter 231 (S640 and S641).

Pixels that have undergone separate processing in the two types of filters are synthesized and output as one image frame (S650). At this time, in order to preserve the object edge existing in the original image, the pixels in the image frame may be synthesized, reconstructed, and output using Equation 3 below.

는 적응 필터(232)에서 출력된 영상 데이터를,

는 저대역 필터(231)에서 출력된 영상 데이터를 의미한다. 그리고, a 및 b는 영상 데이터에 적용되는 가중치로서, 동영상 부호화 장치가 적용되는 기기의 특성, 입력된 영상 프레임의 특성 등에 따라 변경되어 적용될 수 있는 값이다. 즉, 수학식 3은 영상 프레임에 대한 가중치화된 평균을 구하는 식으로서, (a+b)가 '1'이 되는 값들을 적용할 수도 있다. 예를 들어, 비에지 화소나 블록 에지 화소에 대해서는 a를 0으로 설정함으로써, 종래의 디블록킹 기법을 그대로 적용하도록 하고, 객체 에지 화소에 대해서는 b의 값을 0.7 내지 1로 설정하여 줄 수 있다. 수학식 3을 통하여 저대역 필터링 효과를 가지는 디블록킹 효과보다는 원영상의 에지를 강조하는 처리를 함으로써 에지 화소의 일률적인 디블록킹 처리로 인하여 화질이 열화되 는 것을 방지할 수 있다.In equation (3)

The image data output from the adaptive filter 232,

Denotes image data output from the low pass filter 231. In addition, a and b are weights applied to the image data and may be changed and applied according to the characteristics of the device to which the video encoding apparatus is applied, the characteristics of the input image frame, and the like. That is, Equation 3 is a formula for obtaining a weighted average of image frames, and may apply values in which (a + b) becomes '1'. For example, by setting a to 0 for a non-edge pixel or a block edge pixel, the conventional deblocking technique may be applied as it is, and a value of b may be set to 0.7 to 1 for an object edge pixel. Through Equation 3, the processing of emphasizing the edge of the original image rather than the deblocking effect having the low band filtering effect may prevent the image quality from deteriorating due to the uniform deblocking processing of the edge pixels.

In an embodiment of the present invention, all pixel values are set to be input to two types of filters in the filter unit 230, and a separate identification code is added to the object edge pixel to process pixels having the identification code in the adaptive filter 232. As described above, the pixel determined as the object edge pixel by the edge determination unit 220 may be input only to the adaptive filter 232, and other pixels may be input to the low band filter 231 to be processed.

Although the embodiments of the present invention have been described in detail above, the scope of the present invention is not limited thereto, and various modifications and improvements of those skilled in the art using the basic concepts of the present invention defined in the following claims are also provided. It belongs to the scope of rights.

As described above, according to the exemplary embodiment of the present invention, deterioration of the image quality may be prevented by determining the characteristic of the pixel by using the pixel information in the image frame and classifying and selectively filtering the pixel corresponding to the contour region of the object. .

In addition, by effectively eliminating block boundary defects, the natural object edge components of the original image can be selectively preserved, thereby reconstructing the image more clearly.

Claims (15)

A filtering method for removing a blocking phenomenon of an image frame decoded in units of blocks, Extracting pixel information including an edge response size, edge direction information, and edge position information by applying an edge detector to any pixel in the image frame; Classifying the pixel into one of non-edge pixels, object edge pixels, and block edge pixels using the extracted pixel information; And And performing different filtering according to the classification result. The method of claim 1, And the magnitude of the edge response is an absolute value of the response size of applying an edge detector to the pixel. The method of claim 2, The magnitude of the edge response

((i, j): position of any pixel, M (i, j): magnitude of edge response of the pixel, f (i, j): Pixel value of the corresponding pixel, K (m, n): 2D edge detection operator) Blocking of image frames derived from the method. The method of claim 1, And the edge direction information is an inverse tangent of a ratio between a vertical component and a horizontal component obtained by applying an edge detector to the pixel. The method of claim 1, And the edge position is a value indicating a relative position in a block including the pixel as a coordinate. The method according to any one of claims 1 to 5, And the edge detector is a Sobel edge detector or a Canny edge detector. The method according to any one of claims 1 to 5, The pixel classification step is And comparing the magnitude of the edge response of the pixel with a preset threshold and classifying the pixel as an unedge pixel when the threshold is large. The method of claim 7, wherein The pixel classification step is If the edge response is large, it is determined whether the edge direction of the pixel corresponds to 0 ° or 90 °. If the edge direction is not 0 ° or 90 °, the pixel is classified as an object edge pixel. Doing; And When the edge direction corresponds to 0 ° or 90 °, it is determined whether the pixel exists in the block boundary area by using edge position information of the pixel, and if the pixel is located in the block boundary area, the block edge pixel Blocking phenomenon of the image frame further comprising the step of classifying. The method of claim 8, If the edge direction corresponds to 0 ° or 90 °, And determining that the pixel is an object edge pixel if it is determined that the edge position of the pixel does not correspond to a block boundary region. The method of claim 9, The filtering step And a low-band filtering of the pixels classified into non-edge pixels and block edge pixels among the pixels in the decoded image frame. The method of claim 9, The filtering step And removing the pixels classified as object edge pixels among the pixels in the decoded image frame by using a full-band filter or a sharpening filter. The method of claim 9, And adding an identification code to a pixel classified as an object edge pixel among the pixels in the decoded image frame. A pixel information extracting unit extracting the magnitude, edge direction information, and edge position information of an edge response by applying an edge detector to an arbitrary pixel in an image frame decoded in block units; An edge discriminator that classifies the pixel into one of a non-edge pixel, an object edge pixel, and a block edge pixel using the extracted pixel information; And And a filter unit for performing different filtering according to the pixel classification result. The method of claim 13, The filter unit A low band filter for removing high frequency components included in the non-edge pixel and the block edge pixel; And And an adaptive filter for performing filtering to maintain edge information of the object edge pixel. The method of claim 14, And the adaptive filter is selected from a full-band filter or a sharpening filter.

Images