JPH096906A - Method and device for image processing - Google Patents

Abstract

PURPOSE: To handle an agate distinctively from, for example, a main text other than the agate when an image of a document wherein the agate is present is inputted and characters are segmented out of the document and recognized. CONSTITUTION: The document information is inputted (S201), a character image block is extracted from the document image (S202), and it is decided whether or not the extracted character image block is an agate or not (S204); and discrimination information with which the agate can be discriminated is added to an image which is decided to be the agate and they are stored (S206). For recognition processing, a recognition process is performed for character images other than the selected agate according to the discrimination information (S207 and S208).

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image processing method and apparatus capable of cutting out a character area from image information of a document.

[0002]

2. Description of the Related Art Conventionally, a character recognition apparatus, after inputting an image, first takes a projection 801 in the row direction as shown in FIG. 8 and extracts a row like a dotted rectangle 802. A projection 803 is taken in the vertical direction, and an image block for each character is extracted like a solid rectangle 804. Furthermore, in order to correctly extract separated characters and touched characters, the standard character size is calculated from the average width of image blocks in each line or the one with the highest frequency in the width distribution of image blocks. When combining image blocks, if the width is the standard character size, combine them, and if the image block is an integral multiple of the standard character size, divide it into integers and determine the image block for each character. There is also a method of correcting the character image block extracted by the projection of. Also, when it is determined that the image block containing ruby is extracted, the reference ruby separation size is obtained from the average of the heights of the image blocks in each row, or the one with the highest frequency in the height distribution of the image blocks, The main body is an image block whose height exceeds the ruby separation size.
An image block whose height does not exceed the ruby separation size is determined to be ruby, and an image block for each character in which the text and ruby are separated is determined.

After the image block is extracted, an identification calculation is performed to obtain the similarity with the recognition dictionary in which the standard features of each character are stored, and the character with the highest similarity is output as the recognition result. There is.

[0004]

However, in the above-mentioned conventional example, when an image of a document containing ruby is input and the characters are cut out and recognized, the ruby and the text are distinguished after the image block is extracted. There was a defect that there was a line containing only ruby, and ruby was included in the text and output.

An object of the present invention is to prevent ruby and text from being mixed and output even when ruby is extracted from an input document image. In addition, it is necessary to correctly determine the line of only ruby.

[0006]

In order to solve the above-mentioned conventional problems, the present invention extracts an image block of a character from an input document image and determines that the extracted character image is ruby. An image processing method and apparatus are provided.

In order to solve the above-mentioned conventional problems, the present invention preferably stores the extracted character image in a storage means, and the character image determined to be the ruby is ruby. Is added and stored in the storage means.

In order to solve the above conventional problems, the present invention preferably outputs the extracted character image block.

In order to solve the above conventional problems, the present invention preferably controls the output so as not to output the character image judged to be the ruby.

Further, in order to solve the above-mentioned conventional problems, the present invention preferably outputs an image block other than the image block determined to be the ruby among the image blocks of the extracted characters. To control.

In order to solve the above conventional problems, the present invention preferably recognizes the output character image block.

In order to solve the above-mentioned conventional problems, the present invention preferably displays the recognized result on the display means.

In order to solve the above conventional problems, the present invention preferably prints the recognized result on a printing means.

In order to solve the above-mentioned conventional problems, the present invention preferably selects, from the character images stored in the storage means, a character image to which no identification information indicating ruby is added. Control to perform recognition.

In order to solve the above-mentioned conventional problems, the present invention preferably recognizes a character image extracted by the character image extracting means, and a character image judged not to be ruby among the recognized results. The result recognized from is selected and output.

In order to solve the above-mentioned conventional problems, the present invention preferably stores the input document image in a storage means and deletes the character image determined to be the ruby from the storage means. .

In order to solve the above-mentioned conventional problems, the present invention preferably extracts an image block of a line from the input document image in place of the extraction of the character image, and the ruby determination is the extraction. The determined row image is determined to be a ruby row.

In order to solve the above conventional problems, the present invention preferably determines the ruby based on whether or not there is a line before and after the line of interest, and determines that there is no line before and after. When it is determined that the image of the line of interest is not ruby.

In order to solve the above-mentioned conventional problems, the present invention preferably determines the ruby by comparing the line width of the line of interest with the line width of the previous line, and determines the relative to the previous line. When the value is smaller than the reference value, it is determined that the image of the target line is ruby.

In order to solve the above-mentioned conventional problems, the present invention preferably determines the ruby by comparing the line width of the line of interest with the line width of the subsequent line, When the relative value is smaller than the reference value, it is determined that the image of the target line is ruby.

In order to solve the above-mentioned conventional problems, the present invention preferably determines the ruby based on the line spacing between the line of interest and the preceding and following lines, and determines the line spacing between the preceding line and the following line. If the line spacing is larger than the line interval, the image of the line of interest is determined to be ruby.

In order to solve the above conventional problems, the present invention preferably determines the ruby based on the position of the line of interest, and determines that the line of interest is located within a reference range. In this case, it is determined that the line of interest is ruby.

[0023]

【Example】

(First Embodiment) FIG. 1 is a block diagram showing the arrangement of an image processing apparatus for carrying out the present invention. 101 is a ROM
A CPU that controls the entire apparatus according to a control program stored in 102, 102 is a ROM that stores a control program of the apparatus such as the processes shown in the flowcharts described below that the CPU 101 executes, and 103 is a document image or character recognition. , A RAM for storing parameters such as “a” used during processing, 104 is an external storage device such as a magnetic disk, 105 is a display such as a CRT or liquid crystal display, 106 is a keyboard, 10
Reference numeral 7 is a pointing device such as a mouse, and 108 is an image scanner for optically reading an image. 1
09 is an LBP or inkjet printer, 110
Is a communication I / F that controls data communication with the outside via a public line or the like, and enables input of images and output of results. Reference numeral 111 is a data bus used for transferring data between the components.

The processing of this embodiment executed by the image processing apparatus having the configuration shown in FIG. 1 will be described with reference to the flowchart of FIG.

In step S201, a document image optically read by the scanner 108 or the like is input. In S202, image blocks are extracted in character units from the input document image as described in the conventional technique. In S203, information such as the position of the extracted line and the position and size of the character image block cut out from the line rectangle is stored in the RAM 103. Further, in S204, it is determined whether the extracted image block is ruby. An example of the detailed process of the ruby determination method in S204 will be described with reference to the flowchart of FIG.

The ruby determination method described below is S202.
The ruby character image is determined by determining whether or not the line extracted in (3) is composed of only ruby.

In step S301, a line including the image block of interest is obtained, and it is determined whether or not there are lines before and after the line. The determination may be made using the position information of the row rectangle stored in the RAM 103 in S203. If there is no line before and after the line of interest and it is determined that the line is a single line, it can be determined that the line is not a ruby image block (S307). S
In step 302, the width of the line of interest is compared with the value obtained by multiplying the width of the line before it by a predetermined value a of less than 1, and it is checked whether or not the value of the line of interest is smaller. “A” is a parameter which is preset as a reference value for judging the ruby line by comparing the line widths of the preceding and following lines and is stored in the RAM 103, and takes a value of about 0.65 in this embodiment. The line width may be the vertical size of the line rectangle if the document of the input image is written horizontally, and the horizontal size if the document is written vertically. Also, instead of the width of the block cut out as a line rectangle,
It may be the maximum value in the width direction of the row of the image blocks determined to be included in the row. In S303, the same comparison as that performed for the previous line is performed for the line next to the target line. If it is determined in S302 or S303 that the line width of the line of interest is not very small compared to the line width of the previous line or the next line (NO determination), it can be determined that the line of interest is not the ruby image block. (S307).

Next, in S304, the spaces between the preceding and succeeding lines are compared. The position information of the line rectangle stored in the RAM 103 is used to compare the line interval between the target line and the previous line with the line interval between the target line and the next line. If the line of interest is a ruby line, the space between it and the next line should be narrower, so if it is judged that the line space between the previous line and the next line is larger than the line space between the next line and the next ruby line, Although the process proceeds to the determination step, if it is determined to be small (NO determination), it can be determined that the target line is not ruby (S307). Further, the existence range of the line is checked in S305. The existence range is between the left end of the first block and the right end of the last block of the line in horizontal writing. If the line of interest is a ruby line, it does not exceed the existing range of the next line, so the line of interest only needs to exist within the line range of the next line. Note that if there are ruby characters on both ends, the range of the text may be slightly exceeded, so the range of the next line may be widened to the left and right by a predetermined value. In that case, the value may be a standard character.

It is possible to determine that the line of interest remaining at this stage is a ruby line. In S306, the block of interest is set as a ruby image block, and in other cases, it is determined in S307 that it is not a ruby image block. The above-described processing of S301 to S307 is repeated for all the row rectangles extracted in S202 and stored in the RAM 103.

Note that the other image blocks in the line of interest are also ruby image blocks, and there is no need to make the above determination again.
In addition, S301 described in the flowchart of FIG.
The processing steps of ruby determination from S305 to S305 may not be always performed, but may be performed selectively.

When the ruby image block determination is completed by the processing shown in the flowchart of FIG. 3, if the image block determined to be ruby in S204 for each row rectangle in S205, it can be distinguished from other image blocks in S206. The attribute indicating that it is a ruby is added to the image block and stored in the RAM 103.
The processing of 205 and S206 is repeated for all the line rectangles processed in S204.

In S207, among the extracted image blocks, those other than the one determined to be ruby are selected from the RAM 103 based on the attribute added to the image block, and the features are extracted from the image block selected in S208. An identification operation is performed to obtain a similarity by comparing with a character recognition dictionary in the RAM 103 that stores standard characteristics of each character, and in S209, the character having the highest similarity is displayed as the recognition result of the character image block as a result of the identification operation. 105 is displayed.

As described above, after the ruby image block is determined, only the blocks other than the ruby are discriminated and calculated, so that the ruby discrimination calculation processing for which the display of the recognition result is not desired is not performed. , The throughput can be reduced.

Although the above embodiment has been described with the minimum structural requirements for carrying out the present invention, data such as a program for carrying out the processing for carrying out the present invention is externally supplied to a general-purpose computer, for example. It may be provided or stored in the external storage 104 in advance and stored in the RAM 103. Further, it goes without saying that the ruby determination method for carrying out the present invention is not limited to this embodiment. Further, it is easily inferred that the present invention can be implemented in a horizontally written document and a vertically written document.

The recognition result of the document image displayed in S209 is printed by the printer 109 according to the instruction.

(Second Embodiment) The processing of the second embodiment executed by the image processing apparatus having the configuration shown in FIG. 1 will be described with reference to the flowchart of FIG.

In step S401, a document image optically read by the scanner 108 or the like is input. In step S402, image blocks are extracted in character units from the input document image as described in the related art. In S403, the RAM 103 stores information such as the position of the extracted line rectangle and the position and size of the character image block cut out from the line rectangle. Further, in step S404, it is determined whether the extracted image block is ruby by the method described in detail in step S204. In step S405, the image block determined to be ruby in step S404 is identified, and the identified image block is added to the image block with an attribute indicating that it is ruby that can be distinguished from other image blocks in step S406. Stored in the RAM 103.

At S407, the same discrimination operation as at S208 is performed on all the image blocks extracted at S402. In step S408, based on the attribute added when the image block was stored in the RAM 103 among the extracted image blocks, the result of the identification calculation obtained from the image block other than the one determined to be ruby is selected and displayed on the display 105. Yes (S409).

As described above, displaying the recognition results other than those judged as ruby in the ruby image block judgment has an effect of preventing ruby from being mixedly displayed in the text.

(Third Embodiment) The processing of the third embodiment executed by the image processing apparatus having the configuration shown in FIG. 1 will be described with reference to the flowchart of FIG.

In step S501, the document image optically read by the scanner 108 or the like is input. In step S502, image blocks are extracted in character units from the input document image as described in the related art. In step S503, information such as the position of the extracted line rectangle and the position and size of the character image block cut out from the line rectangle is stored in the RAM 103, and in step S504, the same identification operation as in step S208 is performed on all the extracted image blocks. .

Here, it is determined whether or not the extracted image block is ruby by the method described in detail in S204 (S505), and in S506, an identification obtained from an image block other than the one determined to be ruby in S505. Only the calculation result is selected and displayed on the display 105 in S507.

As described above, by determining whether or not it is ruby at the stage of displaying the recognition result, the same result as that of the third embodiment can be obtained without the step of storing the ruby. There is an effect that can be.

(Fourth Embodiment) The processing of the fourth embodiment executed by the image processing apparatus having the configuration shown in FIG. 1 will be described with reference to the flowchart of FIG.

In step S601, a document image optically read by the scanner 108 or the like is input. In step S602, an image block in character units is extracted from the input document image as described in the related art. In step S603, information such as the position of the extracted line rectangle and the position and size of the character image block cut out from the line rectangle is stored in the RAM 103. Further, in step S604, it is determined whether the extracted image block is ruby by the method described in detail in step S204. In S605, the image block determined to be ruby in S604 is selected, and the image of the selected image block is deleted from the information stored in the RAM 103 in S603 (S606). In step S607, the same identification calculation as in step S208 is performed on the image blocks remaining in the RAM 103, and in step S608, the result of the identification calculation is displayed on the display 105.
To be displayed.

As described above, by deleting the image block determined to be ruby from the storage area, it is possible to reduce the used storage area.

(Fifth Embodiment) The processing of the fifth embodiment executed by the image processing apparatus having the configuration shown in FIG. 1 will be described with reference to the flowchart of FIG.

In step S701, a document image optically read by the scanner 108 or the like is input. In step S702, the lines described in the conventional technique are extracted from the input document image. As described in the first embodiment, if the line position is known, it is possible to determine whether or not it is a ruby line. Therefore, in step S703, the ruby line is determined by the method described in detail in step S204. . If it is determined in S704 that the target line is not a ruby line, image blocks are extracted in character units as described in the conventional technique in S705, and information such as the position and size of the character image block extracted in S706 is obtained. RA
It is stored in M103. If the line of interest is determined to be a ruby line in S704, a ruby image block is extracted in S707, and the extracted character image block information and an attribute indicating that the character image block is ruby are stored in the RAM 103 in S708. Remember.

Thereafter, with respect to the character image block stored in the RAM 103, the same identification calculation as in S208 is performed in S709, and if the block of interest is not ruby in S710, it is displayed on the display 105 in S711.

As described above, judging the ruby at the time of row extraction has the effect of reducing the number of ruby judgments.

It should be noted that, in the above embodiment, the present invention can be implemented even if the image block is not extracted or the discrimination operation is not performed for the ruby line.

In the above embodiments, the minimum description for carrying out the present invention has been given.
It is easily inferred that the present invention can be implemented by combining these.

In the first to fifth embodiments described above, the character image block detected as ruby is not output, but an instruction means for outputting ruby together with the text may be further provided. With such means, the character image block detected as ruby is output with the identification information of "ruby" added, and the character image block with the identification information of "ruby" is the character image block of the text. So that it can be handled separately. Then, only when the ruby is also instructed to be output together with the text or separately from the text, the text image block to which the identification information of “ruby” is added is output, and the ruby recognized based on this information is output. The recognition result may be output to the side of the body in an output mode (color is changed, character size is changed, hatching pattern is applied, etc.) that can be distinguished from the body.

[0054]

As described above, according to the present invention,
By extracting an image block of characters from the document image and determining that the extracted character image is ruby,
It is possible to process the ruby and the information other than the ruby separately from the input image in which the ruby and the text are mixed.

As described above, according to the present invention, the extracted character image is stored in the storage means, and the character image determined to be the ruby has the identification information for identifying the ruby. By additionally storing the information in the storage means, it is possible to process the ruby and the information other than the ruby by using the information stored as the ruby.

As described above, according to the present invention, by outputting the extracted character image block,
It is possible to process each character image.

As described above, according to the present invention, by controlling the output so as not to output the character image determined to be the ruby, it becomes easy to process the image excluding the ruby.

As described above, according to the present invention, among the image blocks of the extracted characters, the output control is performed so as to output the image blocks other than the image block determined to be the ruby. It becomes easy to process image blocks other than ruby.

As described above, according to the present invention, by recognizing the output character image block,
A desired character image can be recognized.

As described above, according to the present invention, the recognition result can be confirmed on the display by displaying the recognized result on the display means.

As described above, according to the present invention, it is possible to obtain the recognition result output on the recording medium by printing the recognized result on the printing means.

As described above, according to the present invention, of the character images stored in the storage means, a character image to which no identification information indicating ruby is added is selected for recognition. By controlling, the recognition process can be performed efficiently.

As described above, according to the present invention, the character image extracted by the character image extracting means is recognized, and the result recognized from the character images judged not to be ruby among the recognized results. It is possible to efficiently output the recognition result by selecting and outputting.

As described above, according to the present invention, the input document image is stored in the storage means, and the character image determined to be the ruby is deleted from the storage means, thereby efficiently. Image information that can be processed can be provided.

As described above, according to the present invention, instead of extracting the character image, an image block of a line is extracted from the input document image, and the ruby determination is performed by the extracted line image. By determining that is a ruby line, ruby is determined at the time of line extraction, so that the subsequent processing can be efficiently performed.

As described above, according to the present invention, the ruby determination is made based on whether or not there is a line before and after the line of interest. By determining that the image of the row is not ruby, ruby can be correctly determined based on the positional relationship between the preceding and following rows.

As described above, according to the present invention, the ruby determination is made by comparing the line width of the line of interest with the line width of the previous line, and the relative value to the previous line is the reference value. If it is smaller, it is determined that the image of the target line is ruby, and thus the ruby can be correctly determined based on the line width with the preceding line.

As described above, according to the present invention, the ruby determination is made by comparing the line width of the target line with the line width of the subsequent line, and the relative value to the subsequent line is the reference value. If it is smaller, it is determined that the image of the target line is ruby, and thus ruby can be correctly determined based on the line width of the subsequent line.

As described above, according to the present invention, the ruby determination is performed based on the line spacing between the target line and the lines before and after, and the line spacing between the preceding line and the subsequent line is determined. If it is larger, it is possible to correctly determine ruby based on the line spacing by determining that the image of the target line is ruby.

As described above, according to the present invention, the ruby determination is made based on the position of the line of interest, and when it is determined that the line of interest is located in the reference range, By determining that the line is ruby, ruby can be correctly determined based on the position range of the line.

[Brief description of drawings]

FIG. 1 is a block diagram of an image processing apparatus according to a first embodiment.

FIG. 2 is a flowchart of character recognition processing performed by the image processing apparatus according to the first embodiment.

FIG. 3 is a flowchart of a ruby determination method according to the first embodiment.

FIG. 4 is a flowchart of character recognition processing performed by the image processing apparatus according to the second embodiment.

FIG. 5 is a flowchart of character recognition processing performed by the image processing apparatus according to the third embodiment.

FIG. 6 is a flowchart of character recognition processing performed by the image processing apparatus according to the fourth embodiment.

FIG. 7 is a flowchart of character recognition processing performed by the image processing apparatus according to the fifth embodiment.

FIG. 8 is a diagram illustrating extraction of a character image block in a conventional example.

Claims (34)

[Claims] 1. An image processing method, comprising: extracting an image block of a character from an input document image, and determining that the extracted character image is ruby. 2. The extracted character image is stored in a storage unit, and the character image determined to be the ruby is added with identification information for identifying the ruby and stored in the storage unit. The image processing method according to claim 1, wherein: 3. The image processing method according to claim 1, wherein the extracted character image block is output. 4. The image processing method according to claim 3, wherein the output is controlled so that the character image determined to be the ruby is not output. 5. The image according to claim 3, wherein output is controlled to output an image block other than the image block determined to be the ruby among the image blocks of the extracted characters. Processing method. 6. The image processing method according to claim 3, wherein the output character image block is recognized. 7. The image processing method according to claim 6, wherein the recognized result is displayed on a display unit. 8. The image processing method according to claim 6, wherein the recognized result is printed on a printing unit. 9. The character image stored in the storage means is controlled so as to select and recognize a character image to which no identification information indicating that it is ruby is added. The image processing method described in. 10. The character image extracted by the character image extracting means is recognized, and the recognized result is selected from the character images judged not to be ruby among the recognized results and output is controlled. The image processing method according to claim 1, wherein: 11. The image processing method according to claim 1, wherein the input document image is stored in a storage unit, and the character image determined to be the ruby is deleted from the storage unit. 12. An image block of a line is extracted from the input document image instead of the extraction of the character image, and the ruby determination is to determine that the extracted line image is a ruby line. The image processing method according to claim 1, wherein: 13. The ruby determination is based on whether or not there is a line before and after the target line, and when it is determined that there is no line before and after the target line, it is determined that the image of the target line is not ruby. The image processing method according to claim 1, wherein: 14. The ruby determination is performed by comparing the line width of the line of interest with the line width of the previous line, and when the relative value with respect to the previous line is smaller than the reference value, the image of the line of interest is displayed. The image processing method according to claim 1, wherein is determined to be ruby. 15. The ruby determination is performed by comparing a line width of a target line with a line width of a subsequent line, and when the relative value with the subsequent line is smaller than a reference value, an image of the target line is displayed. The image processing method according to claim 1, wherein is determined to be ruby. 16. The ruby determination is made based on the line spacing between the target line and the lines before and after, and when the line spacing between the previous line and the subsequent line is larger than the line spacing between the subsequent line and the image of the target line. The image processing method according to claim 1, wherein is determined to be ruby. 17. The ruby determination is made based on the position of the line of interest, and when it is determined that the line of interest is located in a reference range, it is determined that the line of interest is ruby. The image processing method according to claim 1, wherein the image processing method is provided. 18. A document image input means for inputting a document image, a character image extracting means for extracting an image block of a character from the document image input by the document image input means, and a character image extracting means for extracting the character image block. An image processing apparatus, comprising: ruby determining means for determining that a character image is ruby. 19. A character image having storage means for storing the character image extracted by the character image extraction means, wherein the character image determined to be ruby by the ruby determination means can be identified as ruby. The information is added and stored in the storage means.
An image processing apparatus according to claim 1. 20. The image processing apparatus according to claim 18, further comprising output means for outputting the character image block extracted by the character image extraction means. 21. The image processing apparatus according to claim 20, further comprising an output control unit that controls the output unit so as not to output the character image determined to be ruby by the ruby determination unit. 22. Output control means for controlling the output means so as to output an image block other than the image block determined to be ruby by the ruby determination means, out of the image blocks of the characters extracted by the character image extraction means. The image processing apparatus according to claim 20, further comprising: 23. The image processing apparatus according to claim 20, further comprising a recognition unit that recognizes the character image block output by the output unit. 24. The image processing apparatus according to claim 23, further comprising display means for displaying a result recognized by the recognition means. 25. The image processing apparatus according to claim 23, further comprising a printing unit that prints a result recognized by the recognition unit. 26. A recognition control means for controlling to select and recognize a character image to which no identification information indicating that it is ruby is added, out of the character images stored in said storage means. The image processing device according to claim 20. 27. A control means for recognizing a character image extracted by the character image extracting means, and controlling to select and output the recognized result from the character images judged not to be ruby among the recognized results. The image processing apparatus according to claim 18, further comprising: 28. A storage means for storing the document image input by the document image input means, wherein the character image determined by the ruby determination means is deleted from the storage means. The image processing apparatus described in (6). 29. A line image extracting means for extracting an image block of a line from the input document image is provided in place of the character image extracting means, and the ruby determining means is a line extracted by the line image extracting means. The image processing apparatus according to claim 18, wherein it is determined that the image is a ruby line. 30. The ruby judging means judges ruby based on whether or not there is a line before and after the line of interest.
The image processing apparatus according to claim 18, wherein when it is determined that there is no line before and after, it is determined that the image of the line of interest is not ruby. 31. The ruby judgment by the ruby judging means is made by comparing the line width of the target line with the line width of the preceding line, and when the relative value with respect to the preceding line is smaller than the reference value, The image processing apparatus according to claim 18, wherein the image of the line of interest is determined to be ruby. 32. The ruby judgment by the ruby judging means is made by comparing the line width of the line of interest with the line width of the line after, and when the relative value with the line after is smaller than the reference, The image processing apparatus according to claim 18, wherein the image of the row is determined to be ruby. 33. The ruby judgment by the ruby judging means judges by the line spacing between the line of interest and the preceding and following lines, and when the line spacing between the preceding line and the subsequent line is greater than the line spacing between the following line. The image processing apparatus according to claim 18, wherein the image of the line of interest is determined to be ruby. 34. The ruby determination unit determines ruby based on the position of the line of interest, and determines that the line of interest is ruby when it is determined that the line of interest is located in a reference range. The image processing apparatus according to claim 18, wherein the image processing apparatus makes a determination.