JP4194301B2 - Character recognition system and character recognition program - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a character recognition system and a character recognition program for performing character recognition of a form.
[0002]
[Prior art]
Conventionally, when a form is read with a scanner and the image is displayed on the display, the part where the characters in the form part are read and the part without the form are also displayed as white images. And the inside of the frame is read by a scanner and passed to character recognition software for character recognition of the form.
[0003]
Also, in order to determine the form origin, marks called three-point marks have to be printed on the upper left, upper right, and lower left of the form.
[0004]
[Problems to be solved by the invention]
For this reason, both the non-printed part of the form and the background part without the form become a white image. Therefore, as described above, the operator surrounds the range corresponding to the form with a frame, There is a problem that the troublesome work of recognizing the characters by reading them with a scanner becomes necessary.
[0005]
In addition, in order to determine the form origin, it is necessary to print marks called three-point marks on the upper left, upper right, and lower left of the form, which causes a problem of form restrictions.
[0006]
In order to solve these problems, the present invention detects a form at each end of the form by changing edge emphasis and contrast when the form is read by an image scanner, and forms the form with a black background outside the form. Automatic image generation, character recognition is performed using the point where the black background form image changes from black to white as a reference point, and form images with a black background outside the form are automatically generated for easy and accurate character recognition. The goal is to achieve it well.
[0007]
[Means for Solving the Problems]
Means for solving the problem will be described with reference to FIG.
[0008]
In FIG. 1, a processing device 1 executes various processes according to a program. Here, an edge enhancement processing means 13, a contrast changing means 14, a form edge detecting means 15, a black background adding means 16, a character recognizing means. 17 or the like.
[0009]
The edge enhancement processing means 13 is for enhancing the edge of the image.
The contrast changing means 14 is for changing the contrast of the image.
[0010]
The black background adding means 16 adds a black background to the outside portion of the form in the image.
[0011]
The character recognition unit 17 performs character recognition using a point as a base point (origin) by changing from a black pixel to a white pixel in the image.
[0012]
The image scanner 3 reads the form 2 and generates an image.
Next, the operation will be described.
[0013]
The image scanner 3 reads the form 3 to generate an image, the edge enhancement processing unit 13 emphasizes the edge in the image, and the form detection unit 15 detects the edge of the form (for example, upper, lower, left, right) from the image. The black background adding means 16 added a black background outside the form image in the image, and the character recognizing means 17 changed from a black pixel to a white pixel in the black background added image. Character recognition is performed using the point as the base point (origin).
[0014]
At this time, when the edge of the form cannot be detected due to edge enhancement, the contrast changing unit 14 repeatedly detects the edge of the form (paper) by increasing or decreasing the contrast of the image and brightness as necessary. The edge is detected.
[0015]
Further, when the form edge detecting means 15 cannot detect any one or two of the top, bottom, left, and right of the form (paper) edge on the image, it can detect the position from the other detectable positions. The position that did not exist is estimated from the size of the form.
[0016]
Therefore, when a form is read by an image scanner, edge enhancement and contrast change are performed to detect or estimate each end portion of the form and automatically generate a form image with the outside of the form as a black background. By performing character recognition using the point where the form image changes from black to white as a reference point, it is possible to automatically generate a form image with the black background on the outside from the form and realize character recognition simply and accurately. Become.
[0017]
DETAILED DESCRIPTION OF THE INVENTION
Next, embodiments and operations of the present invention will be sequentially described in detail with reference to FIGS.
[0018]
FIG. 1 shows a system configuration diagram of the present invention.
In FIG. 1, a processing device 1 executes various processes according to a program, and includes 11 to 18 and the like.
[0019]
The image scanner control means 11 controls the image scanner 3 to read an image from the form 2.
[0020]
The image storage means 12 temporarily stores the form image read from the image scanner 3 by the image scanner control means 11.
[0021]
The edge enhancement processing means 13 enhances the edge of the form image in the image read from the form to facilitate detection of the paper edge (edge) of the form.
[0022]
The contrast changing unit 14 changes the contrast of the paper edge (edge) of the form in the image read from the form, and changes the brightness as necessary to easily detect the paper edge (edge).
[0023]
The form edge detection means 15 detects the position of the paper edge of the form (upper, lower, right, left edge positions) from the image after edge enhancement and contrast change by the edge enhancement processing means 13 and the contrast changing means 14. Is.
[0024]
The black background adding unit 16 generates an image in which a black background is added to the outside portion of the form paper in the image detected by the form edge detecting unit 15.
[0025]
The character recognizing means 17 performs character recognition according to a well-known method, starting from the position where the black pixel is changed to the white pixel in the image with the black background added (see FIG. 5).
[0026]
The recognition result display means 18 displays the recognition result recognized by the character recognition means 17 on the display 4.
[0027]
Form 2 is a form for character recognition.
The image scanner 3 reads an image on the form 2.
[0028]
The display 4 displays the recognition result.
Next, the operation of the configuration of FIG. 1 will be described in detail according to the order of the flowchart of FIG.
[0029]
FIG. 2 shows a flowchart for explaining the operation of the present invention.
In FIG. 2, S1 stores the read image. This stores an image generated by reading the form 2 with the image scanner 3 in the work area.
[0030]
In S2, edge enhancement is performed. At this time, the contrast is changed in S3 to facilitate detection of the paper edge of the form in the image (described later with reference to FIG. 3).
[0031]
In step S4, the form edge is detected. In this case, edge enhancement is performed in S2, contrast is changed in S3, and the positions (edge positions on the top, bottom, left, and right) of the paper edge of the form are detected. When one or two of the top, bottom, left, and right edges of the sheet cannot be detected, the position is estimated from the other detected positions.
[0032]
S5 adds a black background. This generates an image with a black background added to the outside of the form sheet detected in S4.
[0033]
S6 recognizes characters. This is based on the point that changes from a black pixel to a white pixel in the image added with a black background in S5, and performs well-known character recognition (see FIG. 5). The result of character recognition is output as character recognition data.
[0034]
By the above, edge enhancement and contrast change of the image read from the form 2 with the image scanner 3 are detected to detect the edge (edge) of the form, and an image with a black background added to the outside of the detected edge of the form is generated. Characters on the form can be recognized based on the point where the black pixel is changed to the white pixel in the generated image.
[0035]
FIG. 4 shows a detailed flowchart of the present invention.
In FIG. 4, the contrast is initially set in S11. This sets the contrast when reading from a form by the image scanner 3 to a default initial setting value.
[0036]
In step S12, an image is read. This is the contrast of the initial setting value in S11 or the contrast after the contrast value is changed in S21, and the image scanner 3 reads the image of the form 2.
[0037]
In step S13, edge enhancement processing is performed. This performs processing for emphasizing the edge (paper edge) of the form for the image read in S12.
[0038]
S14 detects an upper side, a left side, a right side, and a lower side.
In step S15, it is determined whether the upper side and the left side are detected. In the case of YES, the intersection of the upper side and the left side of the form edge detected in S16 is detected, and the process proceeds to S17. On the other hand, in the case of NO in S15, since the upper side and the left side could not be detected, the contrast value is changed in S21 and S12 and subsequent steps are repeated.
[0039]
In S17, it is determined whether or not the intersection is in the image. This is to determine whether the intersection of the upper side and the left side of the form detected in YES of S15 is in the image, that is, whether the intersection of the upper side and the left side of the form is included in the image read by the image scanner 3. In the case of YES, an image with a black background added (added a black background outside the form in the image) is generated in S18, and character recognition (a point changed from a black pixel to a white pixel in the image is used as a base point in S19). , Well-known character recognition) is performed, and the process proceeds to S20. On the other hand, in the case of NO in S17, it is found that the intersection is not in the image, that is, the image of the form is not included in the image, and the detection position of the edge (edge) of the form is found to be erroneous detection. Therefore, the contrast value is changed in S21, and S12 and subsequent steps are repeated.
[0040]
In S20, it is determined whether a character recognition error has occurred. This is determined as an error as a result of the character recognition in S19. In the case of YES, since it is found that the edge of the form in the image is erroneously detected, the contrast value is changed in S21, and S12 and subsequent steps are repeated. On the other hand, in the case of NO in S20, it is determined that the character recognition error has not occurred and the character recognition has been normally performed, and thus the process ends.
[0041]
As described above, the contrast of reading the image of the form 2 by the image scanner 3 is changed, and the edge of the image read from the form 2 is edge-enhanced to detect the edge of the form from the image, and the intersection of the upper side and the left side is the image. It is determined that the sides of the normal form can be detected when it is inside, and an image with a black background added to the outside of the detected form is generated, and the point that changes from the black pixel to the white pixel on the image is the base point Thus, it is possible to recognize characters accurately and with good accuracy.
[0042]
4 and 5 are explanatory diagrams of the present invention.
In FIG. 4A, an image is read. This takes in the image read from the form 2 by controlling the image scanner 3 shown in FIG. Here, the central rectangular portion is an image read from the form 2, and the outside is an image of a white background read from the reading stand of the image scanner 3 arranged in the outer portion of the form.
[0043]
FIG. 4B shows a state in which the edge is emphasized and the form edge is emphasized. Here, the form edge emphasizes the edge part of the image read from the form 2.
[0044]
FIG. 4C shows a state in which a form edge is detected and a black background is added. Here, an image is shown in which a black background is added to the outside of the form edge detected in FIG.
[0045]
FIG. 5D shows a state in which a position where scanning is switched from black to white is detected from the edge of the image, and the position where the switching is performed is used as a form edge. Here, the positions at which the scan is switched from black to white by scanning from the upper side and the left side are respectively determined as the form edges.
[0046]
FIG. 5E shows a state where a position (form origin) where the upper side and the left side intersect on the form is extracted. The form origin (form base point) is a position where the upper side and the left side of the form intersect on the image. The image origin (image base point) is the upper left position on the image.
[0047]
(F) of FIG. 5 shows a state in which the character position is determined by calculating the position in the X direction and the Y direction from the image origin in accordance with the OCR definition body to determine the character position. This is because, for example, the item name A in the OCR definition defined in FIG. 5G is located at the X coordinate “30”, Y coordinate “20” from the form origin at the X direction size “5”, Y Since it turns out that the character string of the item name A is described in the area of the direction size “1”,
This is converted into a value from the image origin, and the character recognition of the item name A is performed by a well-known method from the region of the converted value from the image origin in FIG.
[0048]
FIG. 5G shows an example of an OCR definition body. Here, the character string described on the form 2 is defined with the X coordinate from the form origin, the Y coordinate, the X direction size from the X coordinate, and the Y direction size from the Y coordinate in association with the item name. ing.
[0049]
(Appendix 1)
In a character recognition system that performs character recognition of forms,
Means for reading an image from a form using an image scanner;
Means for detecting the paper edge of the form with edge enhancement for the read image;
Means for generating a form image in which the outside of the paper edge of the detected form is a black image, and means for recognizing characters based on a position where the black image outside the form image has changed to a white image. Character recognition system characterized by
[0050]
(Appendix 2)
When the edge of the form cannot be detected by edge enhancement, means for detecting the edge of the sheet by repeatedly detecting the edge of the sheet by increasing / decreasing the image contrast and brightness as necessary is provided. The character recognition system according to appendix 1, which is characterized.
[0051]
(Appendix 3)
Means for estimating, based on the size of the form, the position that could not be detected from the other detected positions when any one or two of the top, bottom, left, and right of the paper edge could not be detected; The character recognition system according to Supplementary Note 1 or Supplementary Note 2, which is provided.
[0052]
(Appendix 4)
In a character recognition program that performs character recognition on forms,
Means for reading an image from a form using an image scanner;
Means for detecting the edge of the form by emphasizing the edge of the image read by the image scanner;
Let the computer function as means for generating a form image in which the outside of the paper edge of the detected form is a black image, and means for recognizing characters based on the position where the black image outside the form image has changed to a white image Character recognition program.
[0053]
【The invention's effect】
As described above, according to the present invention, when the form 2 is read by the image scanner 3, edge enhancement and contrast change as necessary are performed to detect or estimate each end portion of the form 2 to thereby detect the form 2. Since a form image having a black background outside is automatically generated and character recognition is performed using a point where the black background form image changes from black to white as a reference point (form origin), form 2 is used. From this, it is possible to automatically generate a form image with the outside as a black background to realize character recognition simply and with good accuracy.
[Brief description of the drawings]
FIG. 1 is a system configuration diagram of the present invention.
FIG. 2 is a flowchart explaining the operation of the present invention.
FIG. 3 is a detailed flowchart of the present invention.
FIG. 4 is an explanatory diagram (part 1) of the present invention.
FIG. 5 is an explanatory diagram (part 2) of the present invention.
[Explanation of symbols]
1: Processing device 11: Image scanner control means 12: Image storage means 13: Edge enhancement processing means 14: Contrast change means 15: Form edge detection means 16: Black background addition means 17: Character recognition means 18: Recognition result display means 2 : Form 3: Image scanner 4: Display

Claims (4)

In a character recognition system that performs character recognition of forms,
Means for reading an image including the entire form from the form using an image scanner;
For the image including the whole of the read form, means for edge detection to detect the upper edge , the left edge , the lower edge , and the right edge of the form;
Means for determining that the intersections of the upper side, the left side, the lower side, and the right side adjacent to the detected paper edge are correct when included in the read image, while determining that it is a false detection when not included;
Means for generating a form image with black pixels on the upper side, left side, lower side, and upper side of the paper edge of the form determined to be correct ;
A character recognition system comprising: character recognition means based on intersections of adjacent ones of the upper side, left side, lower side, and upper side of the paper edge that change from black pixels to white pixels outside the form image. And a means for detecting the paper edge by repeatedly increasing or decreasing either or both of the contrast value and the brightness of the image and detecting the paper edge when it is determined as the erroneous detection. The character recognition system according to claim 1. When any of the upper side, the left side, the lower side, and the right side of the paper edge of the form is not detected, means for estimating the paper edge that could not be detected from the other based on the size of the form The character recognition system according to claim 1 or claim 2, wherein In a character recognition program that performs character recognition on forms,
Means for reading an image including the entire form from the form using an image scanner;
For the image including the whole of the read form, means for edge detection to detect the upper edge , the left edge , the lower edge , and the right edge of the form;
Means for determining that the intersections of the upper side, the left side, the lower side, and the right side adjacent to the detected paper edge are correct when included in the read image, while determining that it is a false detection when not included;
Means for generating a form image with black pixels on the upper side, left side, lower side, and upper side of the paper edge of the form determined to be correct ;
A character recognition program for causing a computer to function as means for recognizing characters based on intersections of adjacent top and left sides, bottom sides, and top sides of a paper edge changing from black pixels to white pixels outside the form image.

Images