JP4129902B2 - Ruled line erasing method, ruled line erasing apparatus, and recording medium - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a ruled line erasing method, a ruled line erasing apparatus, and a recording medium, and is applied to character recognition of documents such as tables and forms including ruled lines and underlines.
[0002]
[Prior art]
In general, when character recognition of a document including ruled lines such as a table or a form is performed, characters inside the frame are often recognized after extracting information on the ruled lines. In that case, the ruled line in contact with the ruled line and the character may be mistakenly extracted as a character. Therefore, only the character image may be extracted after deleting the ruled line from the image. As a ruled line erasing method, a method of erasing ruled line pixels by converting a black run obtained when extracting a ruled line into white pixels is known (Japanese Patent Laid-Open No. 9-179936). However, this method is effective when the character and the ruled line partially touch each other, but in a situation where the character protrudes from the frame and intersects the ruled line, the character There was a problem that some of them were missing. Therefore, a black run histogram is created in order to deal with intersecting characters, and the black run with the highest frequency is considered as a pixel that constitutes a ruled line. By replacing this black run with a white pixel, the character and the ruled line intersect. A method has been proposed for erasing the character portion so as not to be lost even if there is (Japanese Patent Laid-Open No. 2000-82110) .
[0003]
[Problems to be solved by the invention]
However, even with this method, there is a problem that the ruled line may not be erased cleanly if there is noise in the ruled line part or the thickness of the ruled line is not constant due to image fluctuation. Therefore, an object of the present invention is to propose an accurate ruled line erasing method and apparatus and recording medium capable of separating ruled lines and characters even in the presence of generally considered noise and image fluctuations. in thereby reducing erroneous recognition Oite recognition.
[0004]
[Means for Solving the Problems]
According to a first aspect of the present invention, a step of extracting black runs having a predetermined value or more from a binary image including ruled lines, integrating the black runs to extract ruled lines, and black runs constituting the extracted ruled lines are performed. A step of erasing the ruled line by replacing with a white pixel, and extracting a black pixel connected component as a rectangle for the image from which the ruled line has been deleted, and the ruled line area in the rectangle in contact with the erased ruled line area If it is determined that there is an intersection with the character on the ruled line depending on whether or not there is a pair of rectangles facing each other, A black run in the direction perpendicular to the ruled line is extracted from the image area corresponding to the ruled line, a histogram of the length of the black run is generated, and the black run having a length of the histogram is replaced with a white pixel. Step to erase ruled lines When it is determined that the ruled line does not intersect with the character, a black run having a predetermined value or more in a direction parallel to the ruled line is extracted from the image area corresponding to the ruled line, and the black run is integrated. A ruled line erasing method comprising: extracting a ruled line, and erasing the ruled line by replacing black runs constituting the ruled line with white pixels.
[0005]
According to a second aspect of the present invention, there is provided a means for extracting a black run having a predetermined value or more from a binary image including ruled lines, integrating the black runs to extract a ruled line, and a black run constituting the extracted ruled line. A means for erasing ruled lines by replacing with white pixels, and extracting a black pixel connected component as a rectangle for an image from which the ruled lines have been deleted, and the ruled line area in the rectangle in contact with the erased ruled line area If it is determined that there is an intersection with the character on the ruled line, and whether the ruled line has an intersection with the character, depending on whether there is a pair of rectangles facing each other, A black run in the direction perpendicular to the ruled line is extracted from the image area corresponding to the ruled line, a histogram of the length of the black run is generated, and the black run having a length of the histogram is replaced with a white pixel. Means for erasing ruled lines, and sentences on the ruled lines If it is determined that there is no intersection with the ruled line, a black run with a certain value or more in the direction parallel to the ruled line is extracted from the image area corresponding to the ruled line, and the ruled line is extracted by integrating the black runs. And a ruled line erasing apparatus having means for erasing the ruled lines by replacing black runs constituting the ruled lines with white pixels.
[0006]
The invention of claim 3 is characterized by a computer-readable recording medium storing a program for causing a computer to execute each step of the ruled line erasing method of claim 1.
[0007]
DETAILED DESCRIPTION OF THE INVENTION
FIG. 1 is a flowchart for explaining an embodiment of the present invention, FIG. 2 is a device configuration diagram, and FIGS. 3 to 5 are schematic diagrams for explaining the present invention. Hereinafter, an embodiment of the present invention will be described with reference to these drawings.
S1. A binary image input unit 1 such as a scanner reads a document including ruled lines such as a table or a form and stores it in the image memory 2.
S2. A black run longer than a predetermined length is extracted by the black run extraction unit 3 and stored in the black run memory 4. The length determined here is based on a statistically grasped length of a black run that can be regarded as a ruled line.
S3. The black runs are generated by the black run histogram generator 5 and stored in the black run histogram memory 6. The ruled line extraction unit 7 extracts the ruled line as a rectangle and stores it in the ruled line memory 8. FIG. 3 shows a ruled line and the surrounding image. A black run that can be regarded as a ruled line is a part surrounded by a frame near the center.
[0008]
S4. The ruled line erasing unit 9 erases the ruled line in the area recognized as the ruled line by replacing the black run constituting the ruled line with a white pixel (FIG. 4).
S5. In the rectangle extraction unit 10, the black pixel connected component is extracted as a rectangle from the image after the ruled line is deleted (FIG. 5) and stored in the rectangle memory 11. In the example of FIG. 5, five rectangles R1 to R5 are obtained.
S6. The ruled line contact determination unit 12 is opposed to each other when there is at least one pair of rectangles that overlap in the vertical direction among the rectangles in contact with the erased ruled line region (that is, sandwiching the erased ruled line region). If there is a pair of rectangles) , it is determined that the ruled line has an intersection with a character, and the ruled line is erased by a ruled line erasing method A described below. If there is no rectangular pair that matches the condition, the ruled line is erased by the ruled line erasing method B. In the example of FIG. 5, since the pair of R1 and R4 and R2 and R5 matches the condition, the ruled line is erased by the ruled line erase method A corresponding to the intersection.
[0009]
Ruled line removal method A
S11. A black run in a direction orthogonal (crossing) to the ruled line is extracted from the image area corresponding to the ruled line.
S12. A length histogram is generated for this black run. Then, a run corresponding to the thickness of the ruled line appears as a peak in the ruled line region. The other parts are considered to be characters other than the ruled lines (FIG. 3).
S13. Only the black run with the peak length is erased as a ruled line pixel (FIG. 4). Specifically, run data is retained at the time of extraction, and deletion is performed by replacing only the run having a peak with a white pixel. As a result, even if the ruled line and the character cross each other, it is possible to erase only the pixel consisting only of the ruled line without erasing the pixel of the character part.
[0010]
Ruled line elimination method B
When the ruled line is in a state other than the intersection, the ruled line is erased as follows. The states other than the intersection are the following two.
-Ruled lines and characters are not in contact (completely separated).
-The ruled line and the character touch partly (not protruding). Since this method is relatively resistant to noise, it is considered to be effective in cases other than intersections.
S21. A black run in a direction parallel to the ruled line is extracted for the image area corresponding to the ruled line. At this time, the black run is only longer than a length that can be regarded as a predetermined ruled line.
S22. The extracted black runs are integrated and the ruled lines are extracted as rectangles.
S23. The ruled lines are erased by replacing the black runs constituting the extracted ruled lines with white pixels.
[0011]
【The invention's effect】
As described above, according to the present invention, an appropriate ruled line erasing method is selected according to the contact state between the ruled line and the character, and accurate ruled line erasing can be performed even if there are fluctuations due to noise and images. As a result, it is possible to significantly reduce misrecognition of character recognition.
[Brief description of the drawings]
FIG. 1 is a flowchart for explaining an embodiment of the present invention.
FIG. 2 is an apparatus configuration diagram showing an embodiment of a character extraction apparatus according to the present invention.
FIG. 3 is a schematic diagram illustrating an example of a ruled line rectangle.
FIG. 4 is a schematic diagram illustrating an example of ruled line erasure.
FIG. 5 is a schematic diagram showing an example of rectangular extraction.
[Explanation of symbols]
DESCRIPTION OF SYMBOLS 1 Binary image input part 2 Image memory 3 Black run extraction part 4 Black run memory 5 Black run histogram generation part 6 Black run histogram memory 7 Ruled line extraction part 8 Ruled line memory 9 Ruled line deletion part 10 Rectangular extraction part 11 Rectangular memory 12 Ruled line contact Judgment part

Claims (3)

Extracting a black run of a predetermined value or more from a binary image including a ruled line, and integrating the black run to extract a ruled line;
Erasing the ruled line by replacing the black run constituting the extracted ruled line with a white pixel;
For the image from which the ruled lines are erased, the black pixel connected component is extracted as a rectangle, and among the rectangles in contact with the erased ruled line area, there is a pair of rectangles facing each other across the ruled line area. And determining whether the ruled line has an intersection with a character;
When it is determined that the ruled line has an intersection with a character, a black run in a direction orthogonal to the ruled line is extracted from the image area corresponding to the ruled line, and a histogram of the length of the black run is generated. Erasing the ruled line by replacing the black run with the peak of the histogram with a white pixel;
When it is determined that the ruled line does not intersect with the character, a black run having a predetermined value or more in a direction parallel to the ruled line is extracted from the image area corresponding to the ruled line, and the black run is integrated. Extracting the ruled lines and erasing the ruled lines by replacing the black runs constituting the ruled lines with white pixels;
A ruled line erasing method comprising: Means for extracting a black run above a certain value for a binary image including a ruled line, and extracting the ruled line by integrating the black run;
Means for erasing the ruled line by replacing the black run constituting the extracted ruled line with a white pixel;
For the image from which the ruled lines are erased, the black pixel connected component is extracted as a rectangle, and among the rectangles in contact with the erased ruled line area, there is a pair of rectangles facing each other across the ruled line area. Means for determining whether or not the ruled line has an intersection with a character;
When it is determined that the ruled line has an intersection with a character, a black run in a direction orthogonal to the ruled line is extracted from the image area corresponding to the ruled line, and a histogram of the length of the black run is generated. , Means for erasing the ruled line by replacing the black run with the peak of the histogram with a white pixel;
When it is determined that the ruled line does not intersect with the character, a black run having a predetermined value or more in a direction parallel to the ruled line is extracted from the image area corresponding to the ruled line, and the black run is integrated. Means for extracting the ruled line and erasing the ruled line by replacing the black run constituting the ruled line with a white pixel;
A ruled line erasing apparatus comprising: A computer-readable recording medium storing a program for causing a computer to execute the steps of the ruled line erasing method according to claim 1 .

Images