JP2931041B2 - Character recognition method in table - Google Patents

Description

Description: BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for recognizing characters in a table of a document in a character recognition device.

[Conventional technology]

In a character recognition device, a document image is often divided into a character area, an image area such as a photograph or a figure, a table area, and the like, and different processing is performed on each area.

Regarding the table area, the positions of the ruled lines that make up the table are recognized, the connected black image is tracked for the image within the frame surrounded by the ruled line, the circumscribed rectangle of the black pixel connection is obtained, and these are integrated. Character line is extracted and character recognition is performed.

[Problems to be solved by the invention]

Conventionally, however, the character line extraction processing is performed on the assumption that the characters in each frame of the table are printed in either horizontal writing or vertical writing. Therefore, for example, horizontal writing characters as shown in FIG. In the case of a table with both horizontal and vertical character lines, when horizontal writing (vertical writing) is assumed, vertical character lines (horizontal character lines) are not accurately extracted, and as a result, character recognition is performed correctly. There may not be.

SUMMARY OF THE INVENTION It is an object of the present invention to provide an in-table character recognition method capable of accurately extracting both horizontal and vertical character lines in a table and performing character recognition.

[Means for solving the problem]

The present invention relates to an in-table character recognition method for extracting a frame surrounded by line segments in the main scanning direction and the sub-scanning direction from a table area of a document image, extracting a character line in each frame, and recognizing characters. Based on the length of each frame in the main scanning direction and the length in the sub-scanning direction, it is determined whether the character line in each frame is horizontal writing or vertical writing, and the character line in each frame is determined according to the result of this determination. The method is characterized by switching the extraction method.

[Action]

According to the present invention, it is automatically determined whether a character printed in each frame in a table is horizontal writing or vertical writing, and a character line extraction method suitable for the determined direction is applied. Accordingly, even in a table in which horizontal writing frames and vertical writing frames coexist, character lines in each frame are accurately extracted, and thus the accuracy of character extraction and character recognition in each frame is improved.

〔Example〕

FIG. 1 is a block diagram showing an embodiment of the present invention, and FIG. 2 is a flowchart of a process.

A document is read by a binary image input unit 101 such as a scanner, and the binary image is stored in a binary image memory 102 (processing step 201).

For the document image, the table area recognition unit 103 automatically recognizes the table area using a run-length distribution or the like,
Alternatively, an area specified by the operator is recognized as a table area using a mouse or the like, and an image of the table area is stored in the table area image memory 104 (processing step 202).

For the image of this table area, the line segment extraction unit in the main scanning direction
At 105, the black pixels connected in the main scanning direction are traced to extract a line segment in the main scanning direction, and the coordinates of the starting point and the ending point are stored in the main scanning direction line segment coordinate memory 106 (processing step 203). Similarly, in the sub-scanning direction line segment extraction unit 107, the black pixels connected in the sub-scanning direction in the table area image are traced to extract a line segment in the sub-scanning direction, and the coordinates of the start point and the end point are set in the sub-scanning direction line. It is stored in the minute coordinate memory 108 (processing step 204).

Next, the frame recognizing unit 109 refers to the line segment coordinates stored in each of the memories 106 and 108, recognizes a table frame by a combination of the main scanning direction line segment and the sub-scanning direction line segment, and coordinates the frame, for example, a diagonal vertex. Are stored in the frame coordinate memory 110 (processing step 205). In the frame area extraction unit 111, a frame coordinate memory
By referring to the frame coordinates in 115, the image of the frame area is extracted from the table area image memory 104 and stored in the frame area image memory 112 (processing step 206).

Next, the row direction determination unit 113 creates histograms of the lengths in the main scanning direction and the sub-scanning direction for all the frames with reference to the frame coordinate memory 110 (processing steps 207 and 207).
8). Then, it is determined that all the frames having the maximum frequency in the sub-scanning direction have horizontal writing in the row direction (processing step).
209, 210), it is determined that a frame having the same length in the main scanning direction as the length thereof is a vertically written frame (processing step 21).
1, 212), it is determined that the remaining frame is the frame in the row direction having the larger number of frames determined so far (processing step 21).
3). It should be noted that when obtaining histograms in processing steps 207 and 208, a certain width is given to the length in each scanning direction. If they are within the range, it is determined that they match. The obtained row direction information is transmitted to the row image extraction unit 116 via the circumscribed rectangle extraction unit 114.

For example, in the case of the table shown in FIG. 3, the length of the horizontal writing frame in the sub-scanning direction is all the same (or almost the same), so that the frequency is the maximum. Therefore, all horizontal writing frames in this table are determined to be horizontal writing in processing step 210. Further, the length of the frame in which the vertically written character string “data” of this table is printed is substantially the same as the length of the maximum frequency in the sub-scanning direction (the difference is within a certain width). Therefore, in processing step 212, it is determined that the frame is a vertically written frame.

Next, in the circumscribed rectangle extracting unit 114, the frame area image memory 1
12, the black pixels connected to the image in each frame are tracked, the circumscribed rectangle of the black pixel connection is extracted, and the coordinates of the diagonal vertices are stored in the circumscribed rectangle coordinate memory 115 (processing step). 214).

Next, in the row image extracting unit 116, the circumscribed rectangular coordinate memory 1
15, the circumscribed rectangle of the black pixel connection is integrated with each frame area image in the frame area image memory 112 in the row direction determined by the row direction determination unit 113, and thereby the character line image in the frame is obtained. (Character string image) is extracted and stored in the row image memory 117 (processing steps 215 and 216).

In this manner, for each frame, the line direction, that is, whether it is horizontal writing or vertical writing is determined, and the character line image is extracted by a method suitable for the determined line direction. In a mixed table area, a character line image in any line direction can be accurately extracted.

Next, in the character recognition unit 118, a character image is cut out from the character line image of each frame in the line image memory 117, but since the extraction of the character line image in the preceding stage is accurate, this character image cutout can also be accurately performed. it can. Then, the features of the cut-out character image are extracted, matched with the recognition dictionary for recognition, and the recognition result is output to the outside (processing step 217).

〔The invention's effect〕

As described above, according to the present invention, it is automatically determined whether a character line printed in each frame in a table is horizontal writing or vertical writing, and the character line is printed by a method according to the determined direction. Since extraction is performed, even in a table in which horizontal writing frames and vertical writing frames coexist, character lines in each frame can be cut out with high accuracy, and thus the recognition accuracy of characters in the frame can be improved.

[Brief description of the drawings]

FIG. 1 is a block diagram showing an embodiment of the present invention, FIG. 2 is a flowchart of a process, and FIG. 3 is a diagram showing an example of a table in which horizontal writing and vertical writing are mixed. 101: binary image input unit, 102: binary image memory
103 ... Table area recognition unit, 104 ... Table area image memory,
105 main scanning direction line segment extraction unit 106 main scanning direction line segment coordinate memory 107 sub scanning direction line segment extraction unit 108 sub scanning direction line segment coordinate memory 109 frame recognition unit .., 110... Frame coordinate memory, 111... Frame area extraction unit, 112... Frame area image memory, 113... Row direction determination unit, 114. ... Line image extraction unit 117 117 Line image memory 118 Recognition unit.

Claims (1)

(57) [Claims] An in-table character recognition method for extracting a frame surrounded by line segments in a main scanning direction and a sub-scanning direction from a table region of a document image, extracting a character line in each frame, and recognizing characters. Based on the length of each frame in the main scanning direction and the length in the sub-scanning direction, it is determined whether the character line in each frame is horizontal writing or vertical writing, and the character in each frame is determined according to the result of this determination. A method for recognizing characters in a table, wherein a method of extracting rows is switched.