JP2002279344A - Character recognition device and method, and recording medium - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a character recognition device improving a picking precision of characters and improving the efficiency of character recognition processing, by improving the correction precision of the character recognition result in the character recognition process of a colored image. SOLUTION: This character recognition device recognizes the characters, relative to a character image in the colored image and converts them into codes corresponding to the character image. The colored image is divided into multiple regions based on the color information, and when the character image extends over the region obtained by region dividing, the character image is divided at the boundary of the region so as to conduct character recognition.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a character recognizing apparatus, a method thereof, and a recording medium, and more particularly, to character recognition for a color image by using color information and correcting a recognition result. Regarding the technology to be performed.

[0002]

2. Description of the Related Art Character recognition software for identifying a character image in document image data taken into a computer from a scanner or the like and outputting it as a character code has been used very widely in recent years. In addition, due to the price reduction of color scanners and technological advances, the application range of character recognition software is expanding from monochrome only to color. In the case of performing character recognition processing on such a color document, a character area is first selected from a color image by an appropriate method, binarized, and then subjected to character recognition to form a monochrome image.

[0003]

SUMMARY OF THE INVENTION In character recognition,
It is important to determine an area corresponding to one character and to search for an appropriate word string in post-processing. Especially in Japanese, one character is vertical (even in a vertically written document)
May be divided into a plurality of character areas in the background color area, and a correct recognition result cannot be derived unless a correct combination is selected. In addition, characters may be in contact with each other, and in this case, it is more difficult to extract characters.
For example, according to the technology of Japanese Patent No. 2565150,
Character extraction is performed by estimating the standard character width and standard character height using the marginal distribution in the line direction and the marginal distribution in the direction perpendicular to the line. However, in this technique, when a character string is divided into several regions, characters cannot be cut out correctly when adjacent characters belonging to the adjacent regions are close to each other. To solve this, Japanese Patent Application Laid-Open
According to the technology disclosed in Japanese Patent Application Laid-Open No. 11-110060, backgrounds of the same color are integrated, and character recognition is performed on character images in the integrated area. However, according to this technique, no consideration is given to the color of the characters, and even if the background is the same, there is no effect in extracting characters having different colors. In character recognition, it is extremely difficult to completely eliminate erroneous recognition. For this reason, there is a post-processing in which a character string as a result of character recognition is automatically corrected by morphological analysis (Japanese Patent Laid-Open No. Hei 4-31)
No. 5278). However, even if the recognition result for a character image that seems to be correct is corrected based on the recognition result for the incorrect character delimiter,
As a result, there is a danger that it will be degraded. The present invention has been made to solve the above-described problem. In character recognition processing for a color image, the character extraction accuracy is improved, and the character recognition result correction accuracy is improved by improving the character recognition result correction accuracy. Character recognition device to improve efficiency,
An object of the present invention is to provide a character recognition method and a recording medium.

[0004]

SUMMARY OF THE INVENTION In order to solve the above-mentioned problem, a first aspect of the present invention is a character recognition apparatus which performs character recognition on a character image in a color image and converts the character image into a code corresponding to the character image. In the device, an area dividing unit that divides a color image into a plurality of areas based on color information, and when a character image straddles an area obtained by the area dividing unit, the character image is divided by a boundary of the area. A character recognition unit that performs character recognition processing by dividing the character. According to a second aspect of the present invention, in a character recognition device that performs character recognition on a character image of an input image, an area dividing unit that divides the input image into a plurality of areas based on color information; When characters are cut out from the character image region of the input image, if the characters are straddled by the region obtained by the region dividing unit, 1
A character extracting unit configured to extract a character by regarding the character as not being a single character, and a character recognizing unit that recognizes the character extracted by the character extracting unit. Further, according to a third aspect of the present invention, in the character recognition device according to the first or second aspect, the area dividing unit includes a plurality of character recognition units based on color information for a character color of a character image included in the color image. It is characterized in that it is divided into regions. According to a fourth aspect of the present invention, in the character recognition device according to the first, second, or third aspect, the region dividing unit may be configured such that the color parameter of the color image changes at a gradient equal to or greater than a certain value. , And are not integrated as regions of different colors.

According to a fifth aspect of the present invention, there is provided a character recognition method for performing character recognition on a character image of an input image, wherein the input image is divided into a plurality of regions based on color information. And a character extracting step for extracting a character from a character image area of an input image, wherein the character is extracted as a character if the area extends over the area obtained by the area dividing step. And a character recognizing step of recognizing the character extracted in the character extracting step. According to a sixth aspect of the present invention, in the character recognition method according to the fifth aspect, in the area dividing step, the area is divided into a plurality of areas based on color information on a character color of a character image included in the color image. It is characterized by doing so. Claim 7 of the present invention
In the character recognition method according to claim 5 or 6, when the color parameter of the color image changes at a gradient equal to or higher than a certain value, the area dividing step does not integrate the color image as an area of another color. It is characterized by the following. An eighth aspect of the present invention is a computer-readable recording medium storing a program for causing a computer to function as a character recognition device that performs character recognition on a character image of an input image, An area dividing unit that divides the input image into a plurality of areas based on color information, and when cutting out characters from a character image area of the input image,
A character cutout unit that cuts out a character as if it is not one character when straddling the area obtained by the region division unit, and a character recognition unit that recognizes the character cutout by the character cutout unit Wherein a program for character recognition of a character image of a color image is recorded.

A ninth aspect of the present invention is a character recognition device for recognizing a character in a character image in a color image and converting the character image into a code corresponding to the character image. An area dividing unit that divides the input image into a plurality of areas; a character extracting unit that extracts characters from a character image area of an input image; a character recognizing unit that recognizes a character extracted by the character extracting unit; And a recognition result correction unit for correcting the recognition character candidate by linguistically analyzing a character string generated by combining the character candidates recognized in the above, wherein the recognition result correction unit performs a linguistic analysis on the recognition candidate character string. In this case, if the region spans the region obtained by the region dividing unit, the recognition candidate is interpreted as not a word candidate, and the recognition candidate is corrected. According to a tenth aspect of the present invention, there is provided a character recognition method for performing character recognition on a character image in a color image and converting the character into a code corresponding to the character image. An area dividing step of dividing the input image into characters, a character extracting step of extracting characters from a character image area of the input image, a character recognizing step of recognizing the character extracted in the character extracting step, and a character recognizing step. A recognition result correction step of correcting the recognition character candidate by language analysis of the character string generated by combining the character candidates, the recognition result correction step, when performing language analysis on the character string of the recognition candidate, In the case where the region crosses over the region obtained in the region dividing step, the recognition candidate is interpreted as not a word candidate, and the recognition candidate is corrected. Claim 11 of the present invention is:
A computer-readable recording medium storing a program for causing a computer to function as a character recognition device that recognizes a character in a character image in a color image and converts the character into a code corresponding to the character image. Segmentation unit that divides an image into a plurality of regions based on color information, a character segmentation unit that segments characters from a character image region of an input image, and character recognition that recognizes characters segmented by the character segmentation unit. And a recognition result correction unit for correcting a recognition character candidate by language analysis of a character string generated by combining the character candidates recognized by the character recognition unit, and wherein the recognition result correction unit At the time of language analysis of a character string, when a region straddling the region obtained by the region dividing unit is interpreted as not a word candidate, the recognition candidate is corrected. And I characterized by recording a program.

[0007]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The construction and operation of an embodiment of the present invention will be described below in detail with reference to the drawings. Example 1 (1) Configuration of Example 1 FIG. 1 is a block diagram showing a configuration of a character recognition device according to an embodiment of the present invention. The character recognition device according to the first embodiment of the present invention includes an image input unit 10, an original image storage unit 15, an area division unit 20, an area division result storage unit 25, a binarization processing unit 30,
Binary image storage unit 35, character cutout unit 40, character recognition unit 50, recognition dictionary 45, recognition result storage unit 55, output unit 70
It is composed of

The image input unit 10 reads a color image from an image reading device such as a scanner or a file holding an image and registers the color image in the original image storage unit 15. Original image storage unit 15
Is storage means (for example, a storage device such as a magnetic disk) for storing input color original image information (for example, coordinate position information, color information, and the like for each pixel). The color original image is used for processing of the area dividing unit 20, the binarizing unit 30, the output unit 70, and the like. Area division unit 2
0 reads the color image stored in the original image storage unit 15 and performs color clustering and labeling to divide the image area into image areas that can be regarded as the same color. For this purpose, first, the original image is converted into a color distribution in a color space, and the color distribution is divided based on color information (lightness, hue, etc.).
In this operation, the color distribution of the original image is regarded as one cluster, and the cluster is divided until the cluster division reaches a predetermined number of colors or the cluster cannot be divided. Next, the original image is divided into mesh areas, and it is determined which cluster the average value of the color information of this area belongs to, thereby labeling this area. At the time of this labeling process, if the color of the adjacent area is within the allowable value, the same label is repeatedly given, and the integration of the entire image as the same color area is repeated. Even if the colors of the adjacent areas are within the allowable value, these areas are not integrated when the gradient of the color parameter of the original image is equal to or more than a predetermined value. As described above, the clustering and labeling processes are performed on the entire image. However, it is more effective in the present invention to extract a character image and perform the process only on the character image. Information on the area integrated in this way is registered in the area division result storage unit 25 such as a storage device such as a magnetic disk. As shown in FIG. 2, the contents of the area division result storage unit 25 include the start point coordinates (X, Y) and the end point coordinates (X, Y,
Y), the label number of the included region, and information such as the average color of this region.

The binarization processing section 30 binarizes the color image information stored in the original image storage section 15 and stores it in the binarized image storage section 35. The character cutout unit 40 determines a character image area from the image information stored in the binarized image storage unit 35, cuts out a line from the character area, cuts out a character from the cutout line, and Normalization of the size and removal of noise (dirt, etc.) are performed and temporarily stored. For the line segmentation, a threshold value for judging a line segment, a condition for judging other than a character image (rectangular size of pixel exceeding character size, aspect ratio, etc., such as a rectangular size of a noise pixel, a ruled line, etc.), etc. A parameter is set as a parameter, and a line is cut out at a point where the number of black pixels is 0 in the black pixel distribution. Character extraction is performed by searching the target area of the line extracted in this way for a portion having a sufficiently large number of background color (usually white) pixels in the direction orthogonal to the row direction, and dividing the area to circumscribe the character. A series of small areas consisting of rectangles are formed, and consecutive ones of the small areas are integrated with an appropriate standard while allowing overlap, and a candidate for one character area is obtained.
This criterion includes that the aspect ratio is appropriate as a character. When integrating small areas like this,
By referring to the regions registered in the region division result storage unit 25 that can be regarded as the same color, if these small regions extend over regions having different character colors, they are not integrated. This is because one character is expressed in the same color and is hardly divided.

For example, if the character "add" as shown in FIG. 3 is cut out, it is divided into two regions, "power" and "mouth". In order to integrate them as one character, these areas are determined after comparing the interval and the size of the circumscribed rectangle with the size of the circumscribed rectangle for surrounding characters. However, since these two areas are one character "add", they should generally have the same color. Conversely, if the "power" and "mouth" areas have different character colors, they may be cut out as independent characters. As described above, the result of character recognition of a character region in which a character on the left and right sides of a character color boundary has a relatively narrow partial rectangle or a character region cut out across a boundary in the related art is very reliable. Even if you have, you do not have to cut out such erroneous characters,
Even when characters belonging to different groups are in contact with each other, it is possible to create a one-character area by properly cutting off the characters. The character recognizing unit 50 calculates a feature amount of the character extracted and temporarily stored by the character extracting unit 40, and compares the feature amount with the feature amount of the standard pattern in the standard pattern recognition dictionary 45. A predetermined number of candidates having the highest similarity are selected as candidates for recognition results, and the candidate character codes, their similarities, the character types of the candidate characters, and the like are stored in the recognition result storage unit 55 as recognition result data. The recognition dictionary 45 holds information such as a character code for each character and a feature amount of a standard pattern of the character. The recognition result storage unit 55 is a storage unit (for example, a storage device such as a magnetic disk) that stores the recognition result obtained by the character recognition unit 50. The contents are the position of the image to be processed and the processing result. The output unit 70 compares the recognition result data obtained in the recognition result storage unit 55 with the color image in the original image storage unit 15 and outputs the color image to an output device such as a printer, a display, or a file. In such a configuration, when the character recognition device according to the first embodiment is operated, first, the image input unit 10 is activated, and a document specified by the user is read as a color image from an image reading device such as a scanner, or specified. A color image is read from a file holding the stored image, and the image information is registered in the original image storage unit 15.

Next, the area dividing section 20 is activated, reads the color image stored in the original image storing section 15, performs clustering and labeling by color, and divides the image area into image areas that can be regarded as the same color. The registered regions are registered in the region division result storage unit 25.
Here, the clustering process converts the original image into a color distribution in a color space, considers this color distribution as one cluster, and determines whether the cluster division reaches a predetermined number of colors or based on color information (lightness, hue, etc.). The cluster is divided until it cannot be divided. In the labeling process, the original image is divided into mesh regions, and labeling is performed on this region by determining which cluster the average value of the color information of this region belongs to, and the color of an adjacent region is within an allowable value. Sometimes, the same label is repeatedly given, and the integration of the entire image as a region of the same color is repeated. Even if the colors of the adjacent areas are within the allowable value, these areas are not integrated when the gradient of the color parameter of the original image is equal to or more than a predetermined value. The above-described clustering and labeling processing may be such that a character image is extracted and not performed on the entire image, and is performed only on the character image. After the color image is divided into regions, the binarization processing unit 30 is started, the color image information stored in the original image storage unit 15 is binarized, and stored in the binarized image storage unit 35.

Next, the character cutout section 40 is activated, a character image area is determined from the image information stored in the binarized image storage section 35, a line is cut out from the character area, and a line is cut out from the cutout line. A character is cut out, the size of the character portion is normalized, noise (dirt, etc.) is removed, and the character portion is temporarily stored. This character is cut out by searching for a sufficiently large portion of background color pixels in the direction orthogonal to the row direction, dividing the area into a row of small areas consisting of rectangles circumscribing the character, and Are merged on a suitable basis while permitting duplication, and a one-character area candidate is obtained. When the small areas are integrated, the areas that can be regarded as having the same color registered in the area division result storage unit 25 are referred to, and if these small areas span areas having different character colors, they are not integrated. Next, the character recognition unit 50 is activated, calculates a feature amount for the character cut out by the character cutout unit 40, compares the feature amount with the feature amount of the standard pattern in the standard pattern recognition dictionary 45, A predetermined number of candidates of the recognition result are selected from those having the highest similarity, and the candidate character code, the similarity thereof, the character type of the candidate character, and the like are stored in the recognition result storage unit 55 as the recognition result data. Finally, the output means 70
Is started, the recognition result data obtained in the recognition result storage unit 55 is compared with the color image in the original image storage unit 15, and output to an output device such as a printer, a display, or a file. Therefore, by adopting such a configuration, the knowledge that the character color change point in a color document rarely separates one character or one word is used, and the character color change point is a character region or a word boundary. As a result, the character image area can be effectively extracted. As a result, the accuracy of character recognition for a color document can be improved, and the labor required for correction can be reduced.

(2) Processing Flow FIG. 4 is a flowchart for explaining the processing flow of the first embodiment. A document specified by the user is read as a color image from an image reading device such as a scanner, and the image information is registered in the original image storage unit 15 (step S1).
00). Alternatively, when a color image has already been read by the image reading apparatus and a color image has been registered as a file, the color image may be read by designating the file. The color image stored in the original image storage unit 15 is read, and clustering and labeling by color are performed to divide the image region into image regions that can be regarded as the same color, and each of the labeled regions is registered in the region division result storage unit 25. (Step S110). Here, the clustering process converts the original image into a color distribution in a color space, considers the color distribution as one cluster, and determines whether cluster division reaches a predetermined number of colors or based on color information (lightness, hue, etc.). Divide the cluster until it is no longer possible. In the labeling process, the original image is divided into mesh regions, and labeling is performed on this region by determining which cluster the average value of the color information of this region belongs to, and the color of an adjacent region is within an allowable value. Sometimes, the same label is repeatedly given, and the integration of the entire image as a region of the same color is repeated. Even if the colors of the adjacent areas are within the allowable value, when the gradient of the color parameter of the original image is equal to or higher than a predetermined value, these areas are not integrated. The above clustering and labeling process extracts a character image without processing the entire image,
It may be performed only for the character image. The color image information stored in the original image storage unit 15 is binarized,
Storing in the binarized image storage unit 35 (step S12)
0).

A character image area is determined from the image information stored in the binarized image storage unit 35, a line is cut out from the character area, and a character is cut out from the cut line.
Normalization of the size of the character portion and removal of noise (dirt, etc.) are performed and temporarily stored (step S13).
0). This character is cut out by searching for a sufficiently large portion of background color pixels in the direction orthogonal to the row direction, dividing the area into a row of small areas consisting of rectangles circumscribing the character, and Are integrated to obtain one character area candidates. When the small areas are integrated, the areas that can be regarded as having the same color registered in the area division result storage unit 25 are referred to, and if these small areas span areas having different character colors, they are not integrated. A feature amount is calculated for the cut-out character, the feature amount is compared with the feature amount of the standard pattern in the standard pattern recognition dictionary 45, and a predetermined number of candidates having high similarity are set as candidates for a recognition result,
The candidate character code, its similarity, the character type of the candidate character, and the like are stored as recognition result data in the recognition result storage unit 55 (step S140). The recognition result data obtained in the recognition result storage unit 55 is compared with the color image in the original image storage unit 15 and output to an output device such as a printer, a display, or a file (step S150).

<Embodiment 2> FIG. 5 is a block diagram showing a configuration of a character recognition apparatus according to another embodiment of the present invention. The character recognition device according to the second embodiment of the present invention includes an image input unit 1
0, original image storage unit 15, region division unit 20, region division result storage unit 25, binarization processing unit 30, binary image storage unit 3
5, character extraction unit 40, character recognition unit 50, recognition dictionary 4
5, a recognition result storage unit 55, a recognition result correction unit 60, a language dictionary 65, and an output unit 70. Here, the image input unit 10, the original image storage unit 15, the region division unit 20, the region division result storage unit 25, the binarization processing unit 30, the binarized image storage unit 35, the character cutout unit 40, and the character recognition unit 50 , The recognition dictionary 45, the recognition result storage unit 55, and the output unit 70 have the same functions as those of the respective units described in the first embodiment, and thus are omitted. The recognition result correction unit 60 and the language dictionary 65 will be described below. When considering the arrangement of the character candidates stored in the recognition result storage unit 55 output from the character recognition unit 50, the recognition result correction unit 60 generates a plurality of character strings including a combination of adjacent character candidates, Morphological analysis is performed on these character strings using the language dictionary 65. For these character strings, a plurality of word strings are obtained by searching for words registered in the language dictionary 65. Each word string is obtained based on word weight information, word part of speech information and connection cost between parts of speech. By calculating a cost and selecting a predetermined number of word strings from the one with the lowest cost, unnecessary candidate characters are deleted and the contents of the recognition result storage unit 55 are rewritten.

In the second embodiment, when a word registered in the language dictionary 65 is searched for a character string generated from a combination of adjacent character candidates, the same color registered in the area division result storage unit 25 is used. By referring to the area that can be considered, words that span areas having different character colors are not selected. For example, referring to FIG. 6, when the color of the four characters “Home care” in “East welfare office home care” is the same color and different from the other parts, Four characters are matched as words with the language dictionary 65, and only the remaining “East welfare office” is a language dictionary 65
And morphological analysis may be performed. But,
If "home care" is not colored, all possible words of "east care home care" must be matched with the language dictionary 65. For example,
In this example, a word such as “location” which is an incorrect delimiter is also considered. Language dictionary 6
Reference numeral 5 holds a word dictionary having information such as notation, part of speech, and weight for each word, and connection cost information indicating the possibility of connection between parts of speech. By configuring as in the second embodiment,
In a color document, one keyword is often written in the same character color, so searching as one word (candidate) By limiting the range of characters that can be continued while characters of the same color continue, two or more words Is prevented from being mistakenly regarded as one word, and the possibility of obtaining a correct word string is increased.
Thus, the accuracy of character recognition for a color document can be improved, and the labor required for correction can be reduced.

<Example by Computer> Further, the present invention is not limited to the above embodiment.
For example, the character recognition device of FIG. 1 or FIG. 5 can also be realized by the configuration of the computer device 100 as shown in FIG. That is, the input device 1 includes a keyboard, a mouse, a touch panel, a scanner, and the like, and is used for inputting information. The display device 2 displays and outputs various output information, information input from the input device 1, and the like.
A CPU (Central Processing Unit) 3 runs various programs. Memory 4
Holds the program itself, and the program
Information temporarily created when executed by the PU 3 is stored. The storage device 5 includes an original image storage unit 15, an area division result storage unit 25, a binarized image storage unit 35, a recognition dictionary 45, and a recognition result storage unit 5 handled by the character recognition device of the present invention.
5. It holds the language dictionary 65, programs and temporary information at the time of program execution. The medium drive device 6 is used to mount a recording medium storing programs, data, and the like, read them, and store them in the memory 4 or the storage device 5. Further, it may be used for directly inputting / outputting data or executing a program. The bus 8 connects the above components. Each function constituting the character recognition apparatus shown in FIG. 1 or FIG. 5 is programmed and written in advance on a recording medium such as a CD-ROM, and this CD-ROM is stored in a medium such as a CD-ROM drive at each site. These programs are mounted on a computer equipped with a drive device 6 and these programs are stored in the memory 4 or the storage device 5 of each computer.
And executing the same, it is possible to realize functions similar to those of the above-described embodiment.

As a recording medium, a semiconductor medium (for example, ROM, IC memory card, etc.), an optical medium (for example, ROM,
DVD, MO, MD, CD-R, etc.) and magnetic media (for example, magnetic tape, flexible disk, etc.). Further, not only the functions of the above-described embodiment are realized by executing the loaded program, but also the operating system or the like performs part or all of the actual processing based on the instructions of the program,
The case where the function of the above-described embodiment is realized by the processing is also included. Further, when the program for realizing the above-described embodiment is a semiconductor recording medium such as a ROM, the program is directly loaded into the memory 4 and executed not from the medium driving device 6. Further, a program for realizing the functions of the above-described embodiment is loaded into the memory provided on the function expansion board or the function expansion unit, and the CPU or the like provided on the function expansion board or the function expansion unit is actually executed based on the instructions of the program. And a part of the entire process is performed, and the function of the above-described embodiment is realized by the process.

<Operation in Network Environment of the Present Invention> FIG. 8 shows a configuration of a mode in which the present invention is connected to a network and operated. For example, a server 110 holding a character recognition program and a terminal 120 for inputting an image by a user and outputting the result are connected via a network 9. In this case, the server 110 and the user terminal 120
It is composed of the general-purpose computer device shown in FIG. 7 and has a network connection device 7 for connecting to the network 9. Also, this network 9
Is a transmission path for connecting the server 110 and the user terminal 120, and is generally realized by a cable, and TCP / IP is used as a communication protocol. However, the transmission path is not limited to the cable, but may be any of wireless, wired, and broadcast waves as long as the communication protocol between them is the same.
Area Network), WAN (Wide Area Network), Internet, analog telephone network, digital telephone network (ISD)
N: Integral Service Digital Network), PHS (Personal Handy Phone System), mobile phone network, satellite communication network, and the like can be used. User terminal 120
Receives the character recognition program from the server 110, inputs an image, executes the received character recognition program, and outputs the execution result. This has the advantage that the latest character recognition program can always be used. Also, as shown in FIG.
When the terminal 10 and the terminal 120 are connected via a network, a character recognition program for realizing the function of the present invention is stored in a storage device such as a magnetic disk of the server 110 and distributed to the terminal 120 in a form such as download. It is also possible. Further, a character recognition program for realizing the functions of the present invention may be provided by distributing it by broadcast waves.

[0020]

As described above, according to the present invention,
Focusing on the fact that the character color change point in a color document rarely breaks one character or one word, and considers the character color change point as a character region or word boundary,
Language processing in character region extraction and post-processing can be performed effectively. As a result, the accuracy of character recognition for a color document can be improved, and the labor required for correction can be reduced.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a configuration of a first exemplary embodiment of the present invention.

FIG. 2 is a diagram for explaining a data structure of a region division result storage unit.

FIG. 3 is a diagram for explaining character integration of character cutout;

FIG. 4 is a flowchart illustrating a process according to the first exemplary embodiment of the present invention.

FIG. 5 is a block diagram illustrating a configuration of a second exemplary embodiment of the present invention.

FIG. 6 is a diagram for explaining word division during language analysis.

FIG. 7 is a diagram illustrating a hardware configuration when the character recognition device is implemented by a computer.

FIG. 8 is a diagram showing a configuration of a mode in which the present invention is operated in a network.

[Explanation of symbols]

1 input device, 2 display device, 3 CPU, 4 memory, 5 storage device, 6 medium drive device, 7 network connection device, 8 bus, 9 network, 10 image input section, 15 original image storage section, 20 area division section, 25 area division result storage unit, 30 binarization unit, 35 binarization image storage unit, 40 character cutout unit, 45 recognition dictionary, 50
Character recognition unit, 55 recognition result storage unit, 60 recognition result correction unit, 65 language dictionary, 70 output unit, 100 computer device, 110 server, 120 terminal

Claims (11)

[Claims] 1. A character recognition device for recognizing a character in a character image in a color image and converting the character image into a code corresponding to the character image, wherein an area dividing unit divides the color image into a plurality of areas based on color information. A character recognizing unit that performs a character recognition process by dividing the character image at a boundary of the region when the character image straddles the region obtained by the region dividing unit. Character recognition device. 2. A character recognition device for performing character recognition on a character image of an input image, comprising: an area dividing unit configured to divide the input image into a plurality of areas based on color information; When a character is cut out from a character image region, a character cutout unit that cuts out a character as if it is not one character when straddling the region obtained by the region division unit, and a character cutout unit that cuts out the character. A character recognition device comprising: a character recognition unit that recognizes a shifted character. 3. The character recognition device according to claim 1, wherein the region dividing unit divides the character into a plurality of regions based on color information on a character color of a character image included in the color image. A character recognition device characterized in that: 4. The character recognition device according to claim 1, wherein the area dividing unit is configured to output a different color when a color parameter of the color image changes at a gradient equal to or greater than a certain value. A character recognition device characterized in that it is not integrated as an area. 5. A character recognition method for performing character recognition on a character image of an input image, comprising: an area dividing step of dividing the input image into a plurality of areas based on color information; When extracting a character from a character image area, a character extracting step in which a character is regarded as not a single character when the area extends over the area obtained by the area dividing step, and the character is extracted, and the character extracting step A character recognition step of recognizing a deviated character. 6. The character recognition method according to claim 5, wherein in the area dividing step, the area is divided into a plurality of areas based on color information on a character color of a character image included in the color image. Character recognition method to be characterized. 7. The character recognition method according to claim 5, wherein in the area dividing step, when a color parameter of the color image changes at a gradient equal to or more than a certain value, the area is determined to be an area of another color. Character recognition method characterized by not being integrated. 8. A computer-readable recording medium storing a program for causing a computer to function as a character recognition device for performing character recognition on a character image of an input image, the computer-readable recording medium storing a program An area dividing unit that divides the image into a plurality of areas based on information; and a case where a character is cut out from a character image area of an input image, the character is not one character if the area spans the area obtained by the area dividing unit. A character cutout unit that cuts out a character by considering it as a character, and a character recognizing unit that recognizes a character cut out by the character cutout unit, wherein a program for recognizing a character image of a color image is recorded. Computer readable recording medium. 9. A character recognition apparatus for performing character recognition on a character image in a color image and converting the character image into a code corresponding to the character image, wherein the input image is divided into a plurality of regions based on color information. Combining a dividing unit, a character extracting unit that extracts characters from a character image area of an input image, a character recognizing unit that recognizes characters extracted by the character extracting unit, and a character candidate that is recognized by the character recognizing unit. And a recognition result correction unit for correcting a recognition character candidate by language analysis of the generated character string.When the recognition result correction unit performs language analysis on the character string of the recognition candidate, the region division unit A character recognition device, characterized in that when it extends over an obtained area, it is interpreted as not a word candidate and the recognition candidate is corrected. 10. A character recognition method for performing character recognition on a character image in a color image and converting the character image into a code corresponding to the character image, wherein the input image is divided into a plurality of regions based on color information. A dividing step, a character extracting step of extracting a character from a character image area of the input image, a character recognizing step of recognizing the character extracted in the character extracting step, and a character candidate recognized in the character recognizing step. And a recognition result correcting step of correcting a recognition character candidate by language analysis of the generated character string.The recognition result correction step includes: A character recognition method characterized by interpreting that a candidate is not a word candidate when the obtained region is straddled, and correcting the recognition candidate. 11. A computer-readable recording medium storing a program for causing a computer to function as a character recognition device that performs character recognition on a character image in a color image and converts the character into a code corresponding to the character image. An area dividing unit that divides an input image into a plurality of regions based on color information; a character extracting unit that extracts characters from a character image area of the input image; and a character that is extracted by the character extracting unit. A character recognition unit for recognizing, and a recognition result correction unit for correcting the recognition character candidate by language analysis of a character string generated by combining the character candidates recognized by the character recognition unit, When performing a language analysis on the character string of the recognition candidate, if the character string crosses the region obtained by the region dividing unit, the character A computer-readable recording medium characterized by recording a program to be corrected.