JPH0816719A - Character segmenting character recognition method, method and device by using same - Google Patents

Abstract

PURPOSE:To improve the recognition rate of characters by enabling the segmentation of characters even concerning a character string in a column direction when segmenting characters contained in lines in a row direction. CONSTITUTION:An image block 201 is provided with a standard size and becomes a character image block as it is. An image block 202 contains laterally typeset characters and is provided with lateral width (w1) wider than standard character width. A standard block 203 expresses the standard size of the character image block and is calculated from the image block segmented from a character image by a prescribed method. A threshold value T for identifying the image block including the laterally typeset characters is provided by multiplying a prescribed coefficient to lateral width ws of this standard block 203. Then, the lateral width of each segmented image block is compared with the threshold value T, the image block equipped with the lateral width larger than T is defined as the image block including the laterally typeset characters and concerning this block, dividing processing for laterally typeset character string is performed.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a character segmentation method for extracting an image block in character units from a document image and performing character recognition in block units, and a character recognition method and apparatus using the character segmentation method.

[0002]

2. Description of the Related Art Generally, an optical character recognition device (OCR)
Then, an image block for each character is taken out from the optically read document image, the character is cut out, and character recognition is performed for each cut out character. An example of a recognition process performed by a general OCR will be described with reference to FIGS. 7 and 8.

FIG. 8 is a flow chart showing a procedure of character recognition processing by a general OCR. First, step S
At 701, a document image is input using an image scanner or the like. Next, in step S702, an image block is extracted from the input document image.

FIG. 9 is a diagram for explaining a method of extracting an image block used in a general character recognition process. FIG. 9 shows an example of image block extraction of a vertically written document using projection. First, a projection 801 is taken in the vertical direction to extract a line. After extracting the rows, the image block 803 can be extracted by taking a projection 802 on each row in a direction perpendicular to the row direction.
In this state, there is a block 804 containing a plurality of characters due to the contact of two or more characters, or a block 80 separated for a character having a separated part.
There are 5 or so.

Therefore, in step S703, a character-by-character image block (character image block) is created. First,
The standard character height is obtained by taking the average of the heights of the image blocks in each row or extracting the height with the highest frequency from the distribution of the heights of the image blocks. Then, if the height of the image block is an integral multiple of the standard character height, the height is equally divided by the integer and the image blocks are separated into character image blocks. Furthermore, if the height of a plurality of image blocks is the standard character height when combined, the character image blocks are generated by combining those image blocks.
For example, in FIG. 8, an image block 804 is divided into two to generate two character image blocks, and an image block 805 is generated by combining two image blocks to generate one character image block.

As described above, the character is cut out by generating the character image block in steps S702 and S703.
In 4, the identification calculation is performed for each character image block, and the category (character) having the highest degree of similarity is set as the recognition result.

[0007]

However, in the character segmentation by the above method, the horizontal character set included in the vertically written document as shown in FIG.
Like 01, they are taken out as one character image block, so that they cannot be cut out in character units, and there is a problem that they cannot be recognized or cause erroneous recognition.

The present invention has been made in view of such a problem, and when cutting out a character included in a line in a row direction, it is possible to cut out a character string in a column direction. It is an object of the present invention to provide a character segmentation method for improving the recognition rate of, and a character recognition method and device using the same.

[0009]

[Means for Solving the Problems] and

The character recognition apparatus according to the present invention for achieving the above object has the following configuration. That is, a character recognition device for recognizing a character from an image, wherein a row is extracted based on a projection of the image in a row direction, and a character to be recognized is included based on a projection in a column direction orthogonal to the extracted row. Regarding acquisition means for acquiring a rectangular area, determination means for determining whether or not each of the rectangular areas includes a character string in the column direction, and rectangular areas determined by the determination means to include a character string in the column direction A dividing means for dividing a column direction to generate a new rectangular area, and a recognizing means for performing a character recognizing process for each rectangular area obtained by the obtaining means and the dividing means.

It is determined whether or not a column-direction character string composed of a plurality of characters arranged in the column direction exists in the rectangular area cut out by the projection in the row direction and the column direction, and the character string in the column direction is determined. The rectangular area determined to exist is further divided in the column direction. Character recognition processing is executed for each of the rectangular blocks cut out in this manner.

Further, preferably, the acquisition means extracts a row based on a projection of the image in a row direction, cuts out a recognition target character based on a projection in a column direction orthogonal to the extracted row, and performs the recognition. The substantially smallest rectangular area that contains the target character is acquired, and the determination means determines whether or not each of the rectangular areas includes a character string in the column direction based on the size in the column direction. When multiple characters are lined up in the column direction,
This is because the size in the column direction is larger than that of a normal character, and by using this, it is possible to easily identify the rectangular area including the column direction character string.

Further, preferably, the determination means generates a standard column-direction length of the rectangular area based on the average length of the rectangular area in the row direction obtained by the acquisition means, and each of the rectangular areas is generated. Is determined based on the length in the column direction and the standard length in the column direction. This is because the standard size in the column direction can be set based on the size in the line direction of each character, and it is possible to flexibly deal with various document images having different character sizes.

Further, preferably, the dividing means performs a projection in the row direction on the rectangular area determined by the determining means to include a character string in the column direction, thereby cutting out a character to be recognized to form a new rectangle. Generate a region.

Further, a character recognition device according to another structure of the present invention for achieving the above object has, for example, the following structure. That is, a character recognition device for recognizing a character from an image, wherein a row is extracted based on a projection of the image in a row direction, and a character to be recognized is included based on a projection in a column direction orthogonal to the extracted row. Acquisition means for acquiring a rectangular area, first recognition means for performing character recognition by acquiring a similarity with a standard pattern for each rectangular area acquired by the acquisition means, and similarity acquired by the first recognition means A dividing unit that divides a rectangular region whose degree is smaller than a predetermined value in the column direction to obtain the divided rectangular region, and a second recognizing unit that performs character recognition on each of the divided rectangular regions are provided.

Character recognition processing is performed on the rectangular area extracted by the projection in the row direction and the column direction, and when the similarity is less than a predetermined value, the rectangular area is further divided in the column direction to perform the character recognition processing. I do.

[0016] Preferably, the storage device further comprises storage means for storing a character having a shape similar to a character in the column direction, and the dividing means has a degree of similarity acquired by the first recognizing means smaller than a predetermined value. In this case, or when the recognition result by the first recognition means is a character stored in the storage means, the rectangular area is divided in the column direction to obtain the rectangular area. By registering characters that are likely to be erroneously recognized as a horizontal character string, if such a character is recognized, even if the similarity is greater than or equal to a predetermined value, character division in the column direction and character By performing the recognition process, the recognition accuracy is further improved.

Further, preferably, in at least one of the rectangular areas obtained by dividing by the dividing means, when the similarity obtained by the second recognizing means is equal to or less than a predetermined value, the divided rectangle is obtained. It further comprises selection means for selecting a recognition result of the first recognition means for the rectangular area before the area is divided. This is because, if the degree of similarity rather decreases due to further division by the dividing unit, the recognition rate before the division is adopted to prevent the reduction of the recognition rate.

In the present invention, the row direction and the column direction correspond to the vertical direction and the horizontal direction in the case of a vertically written document, respectively. In a horizontally written document, the row direction corresponds to the horizontal direction and the column direction corresponds to the vertical direction.

[0019]

Preferred embodiments of the present invention will be described below with reference to the accompanying drawings.

[Embodiment 1] FIG. 1 is a block diagram showing the configuration of a character recognition apparatus for carrying out the present invention. 101
Is a CPU, which performs various controls in this apparatus according to a control program stored in the ROM 102.
A ROM 102 stores a control program executed by the CPU 101 and various data. The control program that realizes the processing shown in the flow chart described later is also R
It is stored in the OM 102. 103 is a RAM,
It provides a work area when the CPU 101 executes various processes. The RAM 103 also includes an area for storing an optically read document image and the like. A display 104 displays a read image and various recognition results.
A keyboard 105 is used to input various instructions and data. An image scanner 106 optically reads an image.

Next, the operation of the present embodiment executed by the character recognition device having the configuration shown in FIG. 1 will be described.

FIG. 2 is a diagram for explaining the outline of cutting out a character image block in this embodiment. In the figure, 201 is an image block having a standard size, which is a character image block as it is. Reference numeral 202 denotes an image block for horizontally typed characters, which has a width (w1) wider than the standard character width. Reference numeral 203 is a standard block representing the standard size of the character image block, and the scanner 10
It is calculated by a predetermined method from the image block cut out from the character image read from No. 6. This standard block 20
By multiplying the horizontal width ws of 3 by a predetermined coefficient, the threshold value T for identifying the image block including the horizontal typesetting character can be obtained.
Then, the horizontal width of each clipped image block is compared with the threshold value T, and an image block having a horizontal width larger than T is determined to be an image block including a horizontal writing character, and division processing for horizontal writing characters is performed. For example, FIG. 2 shows a state in which the image block 202 has w1> T, and the image block is further projected in the row direction, and as a result, the image block is subdivided into three blocks.

Next, the operation of the first embodiment will be described in more detail with reference to FIGS. 3 and 4.

FIG. 3 is a flow chart showing the character recognition processing by the character recognition device in the first embodiment. First, in step S201, a document image is input using the image scanner 106, and the obtained image data is stored in the RAM 103. At this time, the input image may be displayed on the display 104.

Next, the vertical writing characters are cut out from the vertical writing document, and the result is stored in the RAM 103.
Next, in step S202, the image block is cut out. In the vertical character cutting, as described in the conventional example, in step S202, a vertical projection is taken to extract a row, and a horizontal projection is taken for each row to extract an image block. However, as it is, if there are horizontal characters in a line, the horizontal width of the horizontal characters becomes the width of the line. Therefore, in this example, the margins on the left and right sides of the cut-out image blocks are cut off to correct the width of each image block.

Next, in step S203, a character-by-character image block (character image block) is generated. Here, as described in the related art, each image block is separated and combined to generate an image block for each character.

Subsequent steps S204 to S208
By this process, it is determined whether or not all the character image blocks are horizontal typesetting characters, and the character image blocks determined to be horizontal typesetting characters are further separated.

In step S204, the character image blocks are focused one by one to check whether or not the character image block includes horizontal characters. If it is determined that the character image block of interest includes horizontal typesetting characters, the process proceeds to step S205 to perform a horizontal character cutting process on the character image block. On the other hand, if it is determined that the character image block of interest does not include horizontal typesetting characters, the process proceeds to step S207. The method of determining whether or not the letter is a horizontal typeset character in step S204 will be described later.

In the clipping process for horizontally-written characters, first, in step S205, an image block is taken out. In general, the clipping process for horizontally-written characters has the same procedure as the clipping process for vertically-written characters. That is, the horizontal and horizontal projections are taken for the input image to extract horizontal rows, and then the vertical projections are taken for each row to cut out the image block. However, in the case of this embodiment,
Since the process is for the image block of interest, the process may be started assuming that the row in the horizontal direction has already been extracted.

Next, in step S206, an image block for each character is generated. This can be performed in the same manner as in the case of vertical writing by replacing "height" of the image block in cutting out the character image block in the vertically written character image with "width". However, since the number of horizontally typed characters in a vertically written document is small, the standard character width is often not obtained accurately. Therefore, this process (step S20
6) may be omitted and the image block obtained in step S205 may be used as it is as a character image block.
Then, the above result is stored in the RAM 103.

In step S207, the focused character image block is updated to the next block. Step S20
In 8, it is determined whether or not the investigation in step S204 has been completed for all the character image blocks. If not completed, the process returns to step S204 and the above-described processing is repeated.

On the other hand, if the investigation in step S204 has been completed for all blocks, the process proceeds to step S210. In step S210, identification calculation is performed for each character image block, and the category (character) having the highest similarity is stored as a recognition result in the RAM 103. Here, the recognition result may be displayed on the display 104.

Next, the determination as to whether the character is a horizontally set character in step S204 will be described in more detail with reference to the flowchart of FIG. FIG. 4 is a flowchart showing a processing procedure for determining whether or not a character image block includes horizontal characters.

First, in step S301, the widths of all character image blocks are obtained. Next, in step S302, for example, the average of the widths of the character image blocks in a line is calculated, and the standard character width is calculated. Where punctuation marks, symbols, etc.
If a character having a remarkably small width is included, the accuracy of the standard character width is reduced, so the following processing is performed, for example.

For example, the standard character height H obtained in the generation of the character image block in the previous step S203,
A character image block having an image block width w in the range of mH <w <nH is extracted based on the values m and n (0 <m <n) obtained empirically in advance. Then, the average value is calculated using these extracted character image blocks, and this is set as the standard character width, whereby the accuracy of the standard character width can be improved. Further, the distribution regarding the width of the character image block may be taken, and the representative value of the class having the highest frequency may be set as the standard character width. Here, the values of m and n described above are preferably about m = 0.7 and n = 1.3.

In step S303, the standard character width w thus obtained is used to determine a threshold value T for judging whether or not each character image block includes a horizontally set character. Generally, in a vertically written document, the horizontal width of a character image block including horizontal typesetting characters is larger than the standard character width ws. Therefore, for example, using the value t (t> 1) empirically obtained, the threshold value T = t
By determining xws and comparing the threshold value T with the width of each character image block, it can be determined whether or not each character image block includes a horizontal character string. Here, the value of t is preferably about t = 1.2.

Therefore, in step S304, the width w of each image block is compared with the threshold value T. Here, if w> T, it is determined that the character image block includes horizontal typesetting characters, and the process proceeds to step S305. On the other hand, if w ≦ T, the process proceeds to step S306, and a normal character image block is set.

As described above, according to the first embodiment, by utilizing the fact that the width of the character image block of horizontal typesetting characters in vertical writing is larger than the width of a normal character image block, horizontal typesetting characters are used. It is possible to extract a character image block including a. Further, by performing the character cutting process for horizontal writing on the character image block, it is possible to cut out even horizontal typesetting characters included in the vertically written document in character units.

It is needless to say that the above-mentioned embodiment is a structural example for carrying out the present invention and various applications are possible. For example, it is configured so that the operator can specify a recognition area or perform an operation to correct the recognition result, automatically determine the recognition area without the operator's intervention, or reduce false recognition after this processing. The processing may be added.

Further, the threshold value T may be changed according to the height of each character image block. In the first embodiment, one threshold value T is determined based on the standard character width obtained from the standard character height. However, when a quadruple-width character is mixed in a full-width character, this quadruple-width character is set to the horizontal direction. It may be recognized as a typeface. Therefore, if the threshold value T is a function of the height of the character image block (for example, threshold value T = k × h, where k is a constant and h is the height of the character image block), such a problem can be solved.

A program for performing the processing of the present invention is externally provided to a general-purpose computer, and RAM is provided.
Needless to say, it may be configured to store the control program of the present apparatus.

[Second Embodiment] Next, a second embodiment will be described. In the above-described first embodiment, it is determined whether or not the character image block includes the horizontal typesetting character based on the size of the character image block. In the second embodiment, first, an identification calculation is performed on each character image block, and a character image block in which the degree of similarity exists only in a category smaller than a predetermined value,
Performs processing as horizontal characters. The configuration of the character recognition device according to the second embodiment is similar to that of the first embodiment (FIG. 1), and thus the description thereof is omitted here.

FIG. 5 is a flow chart showing the procedure of character recognition in the second embodiment. Similar to the above-described embodiment, first, a document image is input in step S401, and then step S402.
In step S403, the image block is taken out, and further, in step S403, the image block is made in character units (character image block). Then, in step S404, an identification calculation is performed for each character image block.

Next, attention is paid to each of the character image blocks, and it is determined whether or not to perform character segmentation and identification calculation as horizontal characters for each character image block (step S405). Here, if it is determined that the remarked character image block is to be reprocessed as a horizontal typesetting character, the process proceeds to step S406.
If the re-processing for horizontal typesetting characters is not executed, the process proceeds to step S410. The determination process in step S405 will be described later with reference to the flowchart of FIG.

In step S406, a character cutting process for horizontal writing is performed on the character image block and a new image block is taken out. Further, step S407
At, a character image block is generated from the image block obtained in step S406. However, as described in the first embodiment, since the standard character width cannot be accurately obtained in the generation of the character image block by separating and combining the image blocks in step S407, this step is omitted, and the standard character width is obtained in step S406. The image block may be used as it is as the character image block.

In step S408, step S40
6. In step S407, identification calculation is performed on the newly cut out character image block. And step S
In 409, the character image block before re-character cutting (the character image block cut out in step S402 to step S403) and the character image block after the re-character cutting (character image block cut out in step S406 to step S407) are extracted. The result of comparison is adopted and the one with the higher reliability is adopted as the recognition result. The determination of the reliability in step S409 will be described later with reference to the flowchart of FIG.

In step S410, the focused character image block is advanced. Then, in step S411, it is determined whether or not all the character image blocks have been examined, and if there is an unprocessed character image block, the process returns to step S405.
On the other hand, when the processing is completed for all the character image blocks, this processing is completed.

Next, the determination method in step S405 described above will be described. FIG. 6 is a flowchart showing a procedure for determining whether or not to reprocess each character image block as a horizontal typeface character.

In the identification calculation of step S404, the similarity between all categories and each character image block is calculated, and the category with the maximum similarity is used as the result of the identification calculation. When trying to identify an image in which similar characters do not exist, such as when the character is cut out incorrectly, the degree of similarity is generally lower than when the cutout is performed correctly. In this example, this property is used to determine whether reprocessing is necessary. Here, the threshold value of the degree of similarity for determining whether or not the character cutout reprocessing is necessary is set to V.

First, in step S501, the preset similarity threshold V is compared with the similarity v of the character image block of interest. If v <V, the process advances to step S503 to determine that the character image block is to be processed again. If v ≧ V, step S
Proceed to 502. In step S502, an erroneous recognition list in which characters such as "10" and "Ю" or "00" and "∞" that are likely to be erroneously recognized are stored in advance is referred to.
It is checked whether the recognition result by 404 is included in this misrecognition list. If the recognition result exists in the false recognition list, the process advances to step S503 to determine that the character image block is to be processed again. The misrecognition list is in ROM
102 or RAM 103.

As described above, the character image block to be subjected to the character cutting process is selected again.

Further, in step S409, which of the recognition results before and after the re-character segmentation is to be adopted is determined. This process will be described below. FIG. 7 is a flowchart showing a procedure for determining which of the result before and after the re-processing is adopted for the character image block on which the re-processing of the character cutting (that is, the character cutting for horizontal writing) is performed.

First, in step S601, the re-processing of the character segmentation (steps S406 to S407).
As a result of the identification calculation (step S408) for the image blocks divided by the horizontal writing character segmentation, it is determined whether or not there is a similarity degree smaller than a predetermined threshold value V. If even one with a low degree of similarity exists, the process proceeds to step S602, and the original result before reprocessing (that is, the result obtained in step S404) is set as the final recognition result. On the other hand, if the degree of similarity in all the newly cut out character image blocks is larger than the threshold value V, it is determined that the horizontal character cutting has succeeded, and the result after reprocessing in step S603 (that is, step S603).
The result obtained in 408) is the final result.

As described above, according to the second embodiment,
By using the similarity calculated by the discrimination operation to determine whether or not to perform the process for horizontal writing in vertical writing, it becomes possible to recognize the horizontal writing part included in the vertical writing document. Further, after the processing of the horizontal writing characters in the vertical writing is completed, it is determined whether or not the result is adopted, which has the effect of reducing the erroneous recognition and improving the recognition accuracy.

Further, it is to be understood that the processing for determining whether or not to perform the reprocessing using the size of the character image block, which has been described in the first embodiment, may be implemented by the configuration added to the second embodiment. There is no end.

As described above, in the case of recognizing a vertically written document, it is impossible to correctly perform it until now by determining the horizontal writing character in the vertical writing and performing the processing for horizontal writing on that portion. In addition, it is possible to recognize vertical writing in horizontal writing mode, more accurately perform input work by OCR, and reduce erroneous recognition and correction of unrecognizable characters.

In the above embodiment, the mixture of horizontal typesetting characters in a vertically written document has been described, but it goes without saying that the mixture of vertical typesetting characters in a horizontal writing document can be realized by the same concept as described above.

The present invention may be applied to a system composed of a plurality of devices or an apparatus composed of a single device. Further, it goes without saying that the present invention can also be applied to a case where it is achieved by supplying a program that causes a system or an apparatus to execute the processing defined by the present invention.

[0059]

As described above, according to the present invention,
When a character included in a line in the row direction is cut out, it is possible to cut out a character string in the column direction, and the character recognition rate is improved.

[0060]

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of a character recognition device for carrying out the present invention.

FIG. 2 is a diagram illustrating an outline of clipping a character image block according to the present embodiment.

FIG. 3 is a flowchart showing a character recognition process by the character recognition device in the first embodiment.

FIG. 4 is a flowchart showing a processing procedure for determining whether or not a character image block includes horizontal characters.

FIG. 5 is a flowchart showing a procedure of character recognition in the second embodiment.

FIG. 6 is a flowchart showing a procedure for determining whether or not to re-process each character image block as horizontal typesetting characters.

FIG. 7 is a flowchart showing a procedure for determining which recognition result before and after reprocessing is adopted.

FIG. 8 is a flowchart showing a procedure of character recognition processing by general OCR.

FIG. 9 is a diagram illustrating a method of extracting an image block used in a general character recognition process.

FIG. 10 is a diagram illustrating an example of a vertically written document including horizontally typed characters.

[Explanation of symbols]

 101 CPU 102 ROM 103 RAM 104 Display 105 Keyboard 106 Image Scanner

Claims (10)

[Claims] 1. A character recognition device for recognizing a character from an image, wherein a row is extracted based on a projection of the image in a row direction, and a recognition target is based on a projection in a column direction orthogonal to the extracted row. Acquisition means for acquiring a rectangular area containing characters, determination means for determining whether or not each of the rectangular areas includes a column-direction character string, and the determination means is determined to include a column-direction character string The rectangular area includes division means for dividing the rectangular area in a column direction to generate a new rectangular area, and recognition means for performing character recognition processing for each rectangular area obtained by the acquisition means and the division means. Character recognition device. 2. The acquisition means extracts a row based on a projection in the row direction of the image, cuts out a recognition target character based on a projection in a column direction orthogonal to the extracted row, and extracts the recognition target character. The substantially smallest rectangular area to be included is obtained, and the determination means determines whether or not each of the rectangular areas includes a character string in the column direction based on the size in the column direction. The character recognition device according to claim 1. 3. The determining means generates a standard column-direction length of the rectangular area based on the average row-direction length of the rectangular area obtained by the acquiring means, and the standard column-direction length of the rectangular area is calculated for each of the rectangular areas. The character recognition device according to claim 2, wherein it is determined whether or not a character string in a column direction is included based on the column direction length and the standard column direction length. 4. The division means performs a projection in the row direction on a rectangular area determined to include a character string in the column direction by the determination means to cut out a recognition target character to generate a new rectangular area. The character recognition device according to claim 1, wherein: 5. A character recognition device for recognizing a character from an image, wherein a row is extracted based on a projection of the image in a row direction, and a recognition target is based on a projection in a column direction orthogonal to the extracted row. Acquisition means for acquiring a rectangular area including a character; first recognition means for performing character recognition by acquiring the similarity to a standard pattern for each rectangular area acquired by the acquisition means; and acquisition by the first recognition means A dividing unit that divides a rectangular region whose similarity is smaller than a predetermined value in the column direction to obtain a divided rectangular region, and a second recognizing unit that performs character recognition on each of the divided rectangular regions. A character recognition device characterized by comprising. 6. A storage means for storing characters having a shape similar to a character in a column direction is further provided, wherein the dividing means has a similarity degree smaller than a predetermined value acquired by the first recognition means, or The character recognition according to claim 5, wherein when the recognition result by the first recognition means is a character stored in the storage means, the rectangular area is obtained by dividing the rectangular area in the column direction. apparatus. 7. If at least one of the rectangular areas obtained by the division by the dividing means has a degree of similarity obtained by the second recognizing means that is equal to or less than a predetermined value, the division of the divided rectangular areas is performed. The character recognition device according to claim 5, further comprising a selection unit that selects a recognition result of the first recognition unit for the previous rectangular area. 8. A character recognition method for recognizing a character from an image, wherein a row is extracted based on a projection in a row direction of the image, and a recognition target is based on a projection in a column direction orthogonal to the extracted row. An acquisition step of acquiring a rectangular area containing a character, a determination step of determining whether or not each of the rectangular areas includes a column-direction character string, and the determination step is determined to include a column-direction character string. It is characterized by comprising a dividing step of dividing a rectangular area in the column direction to generate a new rectangular area, and a recognition step of performing character recognition processing for each rectangular area obtained by the acquisition step and the dividing step. Character recognition method. 9. A character recognition method for recognizing a character from an image, wherein a row is extracted based on a projection in a row direction of the image, and a recognition target is based on a projection in a column direction orthogonal to the extracted row. An acquisition step of acquiring a rectangular area containing a character; a first recognition step of performing character recognition by acquiring a similarity with a standard pattern for each rectangular area acquired by the acquisition step; and an acquisition step of the first recognition step. A dividing step of dividing a rectangular area whose similarity is smaller than a predetermined value in the column direction to obtain a divided rectangular area; and obtaining a similarity with a standard pattern for each of the divided rectangular areas. A second recognition step of performing character recognition, the character recognition method comprising: 10. A character cutout method for cutting out a character from an image, wherein a row is extracted based on a projection in a row direction of the image, and a recognition target is based on a projection in a column direction orthogonal to the extracted row. An acquisition step of acquiring a rectangular area containing a character, a determination step of determining whether or not each of the rectangular areas includes a column-direction character string, and the determination step is determined to include a column-direction character string. A character segmentation method comprising: dividing a rectangular area in a column direction to generate a new rectangular area.