JP2002236876A - Analyzing method and analyzer - Google Patents

Abstract

PROBLEM TO BE SOLVED: To execute proper correction processing of a recognition error in character recognition. SOLUTION: An analyzing method analyzes a character string obtained by recognizing characters of a character image read by an image reading means. This method has a dividing process (S707) of dividing the characters at dividable time by discriminating whether or not the respective characters obtained as a result of recognizing the characters can be divided by using character information, a joining process (S708) of making candidates for the character string by respectively using the characters before joining and after joining at joinable time by discriminating whether or not the respective characters after the dividing process can be joined with adjacent characters by using the character information, an analyzing process (S709 and S710) of analyzing the respective candidates for the character string made in the joining process by using word information, and an extracting process (S711) of extracting a proper character string as a result of an analysis in the analyzing process.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an analysis method and an analysis apparatus for correcting a character recognition error on a character recognition result.
And a computer-readable recording medium on which the program for executing the processing is recorded.

[0002]

2. Description of the Related Art Conventionally, an optical character reader (OCR) or (OLCR) has been used as a means for inputting data to a computer or the like.
It has been known. These reading devices recognize the input image data as characters. In general, a process of performing a linguistic analysis process on the recognized character string to correct a character recognition error is generally performed. Here, the linguistic analysis processing performed on the character string as the character recognition result is as follows.
Basically, matching is performed based on the head match with the word dictionary, and the candidate in the matched dictionary or the result of language analysis such as morphological analysis is used as the correct candidate, and the character string of the character recognition result is corrected. A common practice is to do so.

On the other hand, when recognizing an image of a character string input by handwriting, it often occurs that character extraction fails. Regardless of the case where a cell in the area where characters are input is specified, character strings written in the free format input area often have irregular character pitches, and the possibility of character extraction failure is high. Get higher.
In such a character string after character recognition in which character extraction has failed, in many cases, the number of characters of a word constituting the character string does not match the number of characters stored in a word dictionary to be compared. For this reason, in the collation of the character string after the word in which the character segmentation error has occurred, the collation start position is shifted, so that the collation due to the head match with the word dictionary often fails. As a result, for a character string in which a character segmentation error has occurred, there is a problem in that linguistic analysis processing after character recognition does not function sufficiently and correction is not properly performed.

[0004]

SUMMARY OF THE INVENTION The present invention has been made in consideration of the above problems, and has as its object to realize an appropriate recognition error correction process by performing an analysis process as a post-process of character recognition. In particular, an object of the present invention is to make it possible to appropriately correct a character recognition error with respect to a character recognition result in which a character segmentation error has occurred.

[0005]

In order to achieve the above-mentioned object, an analysis method of the present invention for analyzing a character string obtained by character recognition of a character image read by an image reading means is provided. The character information is used to determine whether or not each of the divided characters can be divided.If the character can be divided, a dividing step of dividing the character, and each character after the dividing step can be combined with an adjacent character. Determining whether or not there is a character string using the character information, and when combining is possible, a combining step of creating a character string candidate using the characters before and after combining, respectively, a character string created in the combining step It has an analysis step of analyzing each candidate using word information, and an extraction step of extracting a proper character string as a result of the analysis in the analysis step.

Further, according to another configuration, the analysis method of the present invention for analyzing a character string obtained by character recognition of a character image read by the image reading means, the character obtained as a result of the character recognition is converted into a character string. Determining whether or not the character can be divided using character information, dividing the character if the character can be divided, and creating a character string candidate using the characters before and after the division, respectively; The method includes an analysis step of analyzing each of the character string candidates created in the step using the word information, and an extraction step of extracting an appropriate character string as a result of the analysis in the analysis step.

Further, according to another configuration, the analysis method of the present invention for analyzing a character string obtained by character recognition of a character image read by the image reading means, the character obtained as a result of the character recognition is converted into a character string. Determining whether or not it is possible to combine with adjacent characters by using character information, and when combining is possible, creating a character string candidate using the characters before and after combining, respectively; And an extraction step of extracting a proper character string as a result of the analysis in the analysis step with respect to each of the character string candidates created in step (1).

According to a preferred aspect of the present invention, the character information includes data indicating whether each character can be divided / combined for a plurality of characters. And data indicating candidates after the combination.

According to a preferred aspect of the present invention, the character information includes data of each character in each of the left-right direction and the up-down direction. When the direction of the character string to be input is the horizontal direction, the character information in the horizontal direction is used, and when the direction of the character string obtained by character recognition is the vertical direction, the character information in the vertical direction is used.

According to a preferred aspect of the present invention, the word information includes data on a plurality of words.

Further, according to a preferred aspect of the present invention, the word information includes data relating to the word of the address place name storing the representation of the address / place name, and the data relating to the word of the address place name is prioritized in the analyzing step. The method further includes a step of setting the information to be used.

According to a preferred embodiment of the present invention, the method further includes a display step of displaying the character string extracted in the extraction step.

According to another aspect of the present invention, there is provided an analyzing apparatus for analyzing a character string obtained by character recognition of a character image read by an image reading means. A first storage unit for storing character information including data indicating whether the combination is possible and data indicating candidates after division / combination when division / combination is possible; and data representing a plurality of words. A second storage unit for storing word information; and determining whether each character obtained as a result of character recognition is divisible by using the character information, and dividing the character when the character is divisible. Dividing means, determine whether each character after the division process can be combined with adjacent characters using the character information,
When combining is possible, combining means for creating character string candidates using the characters before and after combining, respectively, and character string candidates created by the combining means are analyzed using the word information. It has an analyzing means and an extracting means for extracting a proper character string as a result of the analysis by the analyzing means.

According to another aspect of the present invention, there is provided an analyzing apparatus for analyzing a character string obtained by character recognition of a character image read by image reading means, wherein each character can be divided and combined for a plurality of characters. Data indicating whether
First storage means for storing character information including data indicating candidates after division and combination when division and combination are possible, and second storage means for storing word information indicating data relating to a plurality of words As a result of character recognition, it is determined whether or not each obtained character can be divided using the character information.If the character can be divided, the character is divided. Dividing means for creating a character string candidate used; analyzing means for analyzing each of the character string candidates created by the dividing means using the word information; Extraction means for extracting

Further, according to another configuration, there is provided an analyzing apparatus for analyzing a character string obtained by character recognition of a character image read by image reading means, wherein each character can be divided and combined for a plurality of characters. Data indicating whether
First storage means for storing character information including data indicating candidates after division and combination when division and combination are possible, and second storage means for storing word information indicating data relating to a plurality of words It is determined whether or not each character obtained as a result of character recognition can be combined with an adjacent character by using the character information, and when combining is possible, the characters before and after combining are used, respectively. Combining means for creating a candidate character string, analyzing means for analyzing each of the character string candidates created by the combining means using the word information, and a result of the analysis by the analyzing means, an appropriate character string. Extracting means for extracting.

According to a preferred aspect of the present invention, the character information includes data of each character in each of the left-right direction and the up-down direction, and the dividing unit and the combining unit include a character obtained by character recognition. When the column direction is the left-right direction, the character information in the left-right direction is used. When the direction of the character string obtained by character recognition is the up-down direction, the character information in the up-down direction is used.

According to a preferred aspect of the present invention, the word information includes data relating to the word of the address place name storing the notation of the address and the place name, and the data relating to the word of the address place name by the analysis means is given priority. And a means for making settings to be used in common.

Further, according to a preferred embodiment of the present invention, the apparatus further comprises display means for displaying the character string extracted by the extraction means.

Further, in order to achieve the above object, a program of the present invention which can be executed by an information processing apparatus has a program code for implementing any one of the above-described analysis methods.

In order to achieve the above object, a program of the present invention which can be executed by a computer device causes the computer device executing the program to function as any one of the analyzers described above.

Further, in order to achieve the above object, a storage medium of the present invention stores the above-mentioned program.

According to another aspect of the present invention, there is provided a storage medium according to any one of the above-described methods, comprising: a plurality of data indicating whether each character can be divided / combined. And character data including data indicating candidates after division / combination when division / combination is possible.

According to the analysis method after character recognition of the present invention,
Character recognition is performed on the read image data, and the resulting character string candidates are compared with the character configuration dictionary to create character string candidates that cover the possibility of character cutout patterns. By performing linguistic analysis processing on the character string candidates, it is possible to correct the character string appropriately even for an erroneously recognized character string in which a character cut error has occurred.

[0024]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Preferred embodiments of the present invention will be described below in detail with reference to the accompanying drawings.

FIG. 1 is a block diagram showing an operation configuration of the entire character recognition processing apparatus according to the embodiment of the present invention.

In FIG. 1, reference numeral 101 denotes an image data input unit for inputting image data using, for example, an optical reading device; and 102, a character is cut out and recognized from the image data input by the image data input unit 101. A character recognizing unit 103 is a character dividing / combining unit. For each character of the character string recognized by the character recognizing unit 102, first, a character dividing process is performed with reference to a character configuration dictionary described later. Create suggestions. Then, the character string candidate created by the division processing is subjected to character combining processing with reference to a character configuration dictionary, thereby creating a character string candidate.

Reference numeral 104 denotes a linguistic processing analysis unit which applies linguistic processing such as morphological analysis to the character string candidates created by the character division / combination unit 103 to evaluate the validity of the character string candidates. , Create the right candidate. A recognition result display unit 105 outputs a proper character string candidate created by the linguistic processing analysis unit 104 as a proper candidate of a character recognition result by the character recognition unit 102, and displays the recognition result display unit. The character composition dictionary 107 stores information on whether the character can be divided into characters and what characters can be combined with each other to form one character. It is a word dictionary composed of words, notations, readings, grammar information, and the like. Character composition dictionary 106 and word dictionary 1
07 will be described later with reference to FIGS. 4 and 5, respectively.

Reference numeral 108 denotes a storage unit which stores the processing procedure and processing state of the present invention and is also used as a temporary storage such as a working buffer.
8 is a control unit that calls / executes the processing procedure stored in 8, draws characters on the display unit 110, and controls the operation of the entire apparatus in response to a signal from the instruction unit 111. An instruction unit, which is represented by a keyboard and a pointing device such as a mouse, enables input of characters, execution instruction, cancellation instruction, cancellation instruction, and the like.

FIG. 2 is a block diagram showing an example of a hardware configuration of the character recognition processing device according to the embodiment of the present invention.

In FIG. 2, reference numeral 201 denotes a CPU (central control unit); 202, a ROM (read only memory);
Reference numeral 3 denotes a RAM (random read / write memory), reference numeral 204 denotes an FD (floppy (registered trademark) disk) drive,
Reference numeral 05 denotes an FD, 206 denotes an HD (hard disk) drive, 207 denotes a keyboard, 208 denotes a mouse, 209 denotes a display, 210 denotes a printer, 211 denotes an image reader, and 212 denotes a control bus. It consists of various devices.

The display section 110 shown in the operation configuration diagram of FIG. 1 corresponds to the display 209, the instruction section 111 corresponds to the keyboard 207 and the mouse 208, and the storage section 108 has the RA.
The control unit 109 corresponds to the CPU 201, and the image data input unit 101 corresponds to the image reader 211. Other than the above, the dictionary and each part shown in FIG.
D or HD, and is stored in the RAM 20
3 is configured.

FIG. 3 shows that a program or the like is supplied from a medium to a computer system, and 300 is a computer system.

FIG. 4 shows an example of the character configuration dictionary 106 referred to in the character division / combination process performed by the character division / combination unit 103. For each character corresponding to a heading character, a set of characters that can be divided, and 1
Contains a set of characters that can be composed of characters.
(A) is an example of a character configuration dictionary that can be divided and combined in the left-right direction, and is used for a character recognition result of a horizontally input character. (B) is an example of a character configuration dictionary that can be vertically divided and combined, and is used for a character recognition result of a vertically input character.

An example of the case (A) is as follows.
The characters "tree" are "sword", "to", "ge",
Combined with characters such as "strength", "on", "age", "ko", "dim", and "junk", "paku", "desk", "junki", "iri",
This indicates that it is possible to configure characters such as “wood”, “robot”, and “village”. In addition, the character "go" can be divided into three types of characters, such as "shin" and "main", "te" and "main", or "dry" and "main". Is shown. The combination of characters stored here stores a combination of characters expected as a result of character recognition, and includes combinations that cannot actually constitute characters. For example, for the character "O", three sets of characters that can be divided ((shin, lord), (te, lord), and (dry, lord) are registered. Even if they are combined, the characters do not actually become "outgoing." The name of "shin", which is the radical of "go", is registered in consideration of the possibility that characters "te" and "dry" are erroneously recognized as a result of character recognition.

FIG. 4B is an example of a character configuration dictionary that can be divided and combined in the vertical direction. The heading character "I" is the next character (downward in the vertical text) "Ge", "Father",
Combined with "c" and "width", "high", "interchange",
It is possible to compose the characters "six" and "city". Also, "three" is divided in the up-down direction, (1, 2),
This indicates that the character can be divided into three combinations of (2, 1) and (1, 1, 1).

FIG. 5 is an example of a word dictionary referred to in the linguistic analysis processing performed by the linguistic analysis unit 104, and stores information such as word notation, reading, part of speech, and connection information. . The connection information shown here indicates what part of speech can be connected after the target word. For example, "arcade", which is a noun, indicates that a phrase can be formed by connecting it to a word, which is a case particle at the end. This dictionary assumes a case where morphological analysis is performed as a language processing analysis process.

FIG. 6 shows an example of character string candidates stored in the buffer during the processing. (A) shows an example of the content of the character string of the input image, and (B) shows the character recognition result.
(C) is a character string candidate created after performing character division processing using the character composition dictionary 106. (D) is a character string candidate created after performing a character combining process using the character configuration dictionary 106. (E) is a proper character string candidate extracted as a result of performing the linguistic correction processing by the linguistic analysis unit 104.

FIG. 7 is a flowchart showing the operation of the character recognition process according to the present embodiment. The processing procedure will be described below in order based on the flowchart.

First, a program is loaded in step S701. This is a process of loading the character recognition processing program of the present embodiment into the RAM from the external storage device storing the program. 2 and 3, the character recognition processing program is loaded from the FD 205 and executed as an external storage device.
After that, the program may be temporarily stored (installed) in the HD (hard disk) and loaded from the HD.

In step S702, an image for character recognition is input. For example, image data is loaded into the storage unit 108 from an apparatus such as the image reader 211 in FIG.
Thus, a character image as shown in FIG. 6A is read.

Next, in step S703, the storage unit 1
A character recognition process is performed on the image data captured in step 08. Thereby, for example, a recognition result as shown in FIG. 6B is obtained. Various methods have been devised for character recognition. However, the present invention is not limited to the character recognition method, and has no direct relation to the essence of the present invention, so that the description is omitted here. Generally, character recognition is not reliable,
Since there is a possibility of recognition error, a plurality of candidate characters are generally output in the character recognition process, but FIG. 6B shows only one candidate for simplification. This recognition result indicates a case where the character extraction has failed, and the character “horizontal” of “Yokohama” has been divided into two characters of “tree” and “yellow” and recognized.

In step S704, it is determined which of the character configuration dictionaries divided and combined in the horizontal direction or the vertical direction is to be used. If the input image is vertical text,
In step S705, a character configuration dictionary that can be divided and combined in the vertical direction is used. If the input image is a horizontal writing character, a character configuration dictionary that can be divided and combined in the horizontal direction is used in step S706. To do. Here, since the input image is written horizontally, a character configuration dictionary that can be divided and combined in the left-right direction is used in step S706, and the following steps show a specific example in that case.

In step S707, character division candidates are created. For example, referring to the character configuration dictionary 106 shown in FIG. 4A for each character of the character recognition result as shown in FIG. 6B, the character “Hama” is (Shi, Hyo), (Shi, Hyo). 2
Can be divided into two combinations of characters, the characters "Zhang"
Can be divided into (bow, long) and (ro, long). As a result, four character string candidates are created as shown in FIG.

Step S708 is character combination candidate creation processing. For each character of the four candidates in FIG. 6C created as a result of the character division candidate creation processing, the character configuration dictionary in FIG. 4 is referred to. In the example shown in FIG. 6 (C), “I will go to the Mokuhoshi Shi soldier,” the adjacent characters, that is, (Thu,
(Yellow), (yellow, si), (si, soldier), (soldier, ni), (ni,
De), (de, bow), (bow, long), (long, su), (su, ru)
To see if they can be combined into one character. As a result, (tree, yellow) becomes one character "horizontal", (shi,
It can be seen that (Hei) can be “Hama” and (bow, chief) can be “Zhang”, and as shown in FIG. ”,“ I ’m going to head to Kiyohama ”,
"I will go to Yokoshihei", "I will go to Yokohama", "I will travel to Mikohoshi", "I will travel to Kikohama", "I will travel to Yokoshihei", "I will travel to Yokohama" Is created. Similarly, when a character combination candidate creation process is performed on the other division candidates in FIG. 6C, a character string candidate indicated by the combination candidate in FIG. 6C is created, and a total of 32 types of candidates are obtained. Created.

Step S709 is performed in step S704,
The number of candidates (32 in the example of FIG. 6) is held in order to perform linguistic analysis processing for the number of candidates created in step S705.

In step S710, linguistic analysis processing is performed. Here, morphological analysis is performed to evaluate the validity of the character string candidate. By referring to the word dictionary 107 in FIG. 5, character string candidates are decomposed into words (morphemes), phrases are assembled, and it is evaluated whether or not they are grammatically matched as sentences. Then, it is determined that the character string that matches the sentence is a proper candidate. When morphological analysis is performed on the character string candidates shown in FIG. 6D, “Yokohama” (noun) + “ni” (case particle) + “business trip” (noun) + to (sa line inflected verb) Evaluates to be a grammatically correct string.

In step S711, step S710
The candidate character string evaluated as appropriate in the linguistic analysis processing performed in (1) is output. Here, "Travel to Yokohama" is output as a correct answer candidate. In other words, for the character string created as a result of character recognition,
“Travel to Yokohama” is output as the correction result.

(Modification) In the above embodiment, in the candidate creation process using the character configuration dictionary, the combination candidate creation process (step S705) is performed following the division candidate creation process (step S704 in FIG. 7). Creating character string candidates. However, only one of the two processes may be performed. This is because, in character recognition, a case where one character is divided into two characters and recognized is compared with a case where two characters are recognized as one character,
This is because either frequency may increase depending on the characteristics of the input image and the character recognition processing. In that case, by performing only one of the division candidate creation process and the combination candidate creation process, the number of candidates in the processing process can be reduced, and the processing speed can be improved.

Further, in the linguistic analysis processing performed in step S710 of FIG. 7, the input character string may be limited to the address place name. In this case, an address place name dictionary which stores address place name data and expresses a hierarchical structure of addresses is used in the word dictionary 107 shown in FIG. Then, in the language processing analysis processing in step S710 of FIG. 7, a collation processing with an address dictionary is performed. These dictionaries and processes are generally omitted from description because some methods have been proposed.
By limiting the character string to be subjected to the analysis processing after the character recognition to the address and place name, the character string candidates created in steps S707 and S708 in FIG. 7 are limited, so that the accuracy and processing speed of the analysis processing are improved. be able to.

[0050]

[Other Embodiments] Even if the present invention is applied to a system composed of a plurality of devices (for example, a host computer, an interface device, a scanner, a video camera, etc.), an apparatus composed of one device (for example, Machine, facsimile machine, etc.).

The object of the present invention is, as described above,
A storage medium (or recording medium) that records software program codes for realizing the functions of the embodiments is supplied to a system or an apparatus, and a computer (or CPU or MPU) of the system or the apparatus stores the program stored in the storage medium. This can be achieved by reading and executing the code. In this case, the program code itself read from the storage medium implements the functions of the above-described embodiment, and the storage medium storing the program code constitutes the present invention. When the computer executes the readout program codes, not only the functions of the above-described embodiments are realized, but also an operating system (OS) running on the computer based on the instructions of the program codes. It goes without saying that a case where some or all of the actual processing is performed and the functions of the above-described embodiments are realized by the processing is also included. Here, as a storage medium for storing the program code, in addition to the floppy disk described above, for example, a hard disk, a ROM,
RAM, magnetic tape, nonvolatile memory card, CD-
ROMs, CD-Rs, DVDs, optical disks, magneto-optical disks, MOs, and the like are conceivable.

Further, after the program code read from the storage medium is written into a memory provided in a function expansion card inserted into the computer or a function expansion unit connected to the computer, based on the instruction of the program code. , The CPU provided in the function expansion card or the function expansion unit performs part or all of the actual processing,
It goes without saying that a case where the function of the above-described embodiment is realized by the processing is also included.

When the present invention is applied to the storage medium, the storage medium stores program codes corresponding to the above-described flowchart shown in FIG.

[0054]

As described above, according to the present invention, appropriate analysis error correction processing can be realized by performing analysis processing as post-processing of character recognition. In particular, it is possible to appropriately correct a character recognition error with respect to a character recognition result in which a character segmentation error has occurred.

[Brief description of the drawings]

FIG. 1 is a block diagram showing an operation configuration of a character recognition processing device according to an embodiment of the present invention.

FIG. 2 is a block diagram illustrating a hardware configuration of a character recognition processing device according to the embodiment of the present invention.

FIG. 3 is a diagram showing that a program or the like is supplied from a medium to a computer system.

FIG. 4 is a diagram showing an example of the contents of a character configuration dictionary according to the embodiment of the present invention.

FIG. 5 is a diagram showing an example of the contents of a language analysis word dictionary according to the embodiment of the present invention.

FIG. 6 is a diagram for describing analysis processing after character recognition according to the embodiment of the present invention.

FIG. 7 is a flowchart illustrating an operation of a character recognition process according to the present embodiment.

[Description of Signs] 101 Image data input unit 102 Character recognition unit 103 Character division / combination unit 104 Linguistic analysis unit 105 Recognition result display unit 106 Character composition dictionary 107 Word dictionary 108 Storage unit 109 Control unit 110 Display unit 111 Instruction unit 201 CPU (Central Control Unit) 202 ROM (Read Only Memory) 203 RAM (Read and Write Memory as needed) 204 FD (Floppy Disk) Drive 205 FD (Floppy Disk) 206 HD (Hard Disk) Drive 207 Keyboard 208 Mouse 209 Display 210 Printer 211 Image reader 212 control bus

Claims (19)

[Claims] An analysis method for analyzing a character string obtained by character recognition of a character image read by an image reading means, the method comprising: determining whether each character obtained as a result of character recognition can be divided; Determining using information, and dividing the character when the character can be divided, and using the character information to determine whether each character after the dividing step can be combined with an adjacent character. A combining step of creating character string candidates using characters before and after combining when the combining is possible, and analyzing each of the character string candidates created in the combining step using word information. An analysis method, comprising: an analysis step; and an extraction step of extracting an appropriate character string as a result of the analysis in the analysis step. 2. An analysis method for analyzing a character string obtained by character recognition of a character image read by an image reading means, the method comprising: determining whether each character obtained as a result of character recognition can be divided; Determining using the information, dividing the character if the character can be divided, and creating a character string candidate using the characters before and after the division, respectively, a character string candidate created in the dividing step An analysis method, comprising: an analysis step of performing analysis using word information; and an extraction step of extracting a proper character string as a result of the analysis in the analysis step. 3. An analyzing method for analyzing a character string obtained by character recognition of a character image read by an image reading means, wherein each character obtained as a result of character recognition can be combined with an adjacent character. Whether or not to determine whether or not the character string candidate using the character information, if the combination is possible, a combination step of creating a character string candidate using the characters before and after combination, respectively, character string candidates created in the combination step An analysis method characterized by comprising: an analysis step of performing analysis using word information; and an extraction step of extracting a proper character string as a result of the analysis in the analysis step. 4. The character information includes: data indicating whether each character can be divided / combined for a plurality of characters;
The analysis method according to any one of claims 1 to 3, further comprising, when combination is possible, data indicating candidates after division / combination. 5. The character information includes data of each character in each of a horizontal direction and a vertical direction, and in the dividing step and / or the combining step, a direction of a character string obtained by character recognition is a horizontal direction. 5. The analysis method according to claim 4, wherein in the case, the character information in the horizontal direction is used, and when the direction of the character string obtained by character recognition is the vertical direction, the character information in the vertical direction is used. 6. The analysis method according to claim 1, wherein the word information includes data on a plurality of words. 7. The word information includes data relating to the word of the address place name which stores the notation of the address / place name, and further comprises a step of setting the data relating to the word of the address place name in the analysis step to be preferentially used. The analysis method according to claim 1, wherein: 8. The analysis method according to claim 1, further comprising a display step of displaying the character string extracted in the extraction step. 9. An analyzing apparatus for analyzing a character string obtained by character recognition of a character image read by an image reading means, comprising: data indicating whether each character can be divided / combined for a plurality of characters; A first storage unit for storing character information including data indicating candidates after division / combination when division / combination is possible; and a second storage unit for storing word information representing data relating to a plurality of words.
Storage means for determining whether or not each character obtained as a result of character recognition is divisible, using the character information, and dividing the character if it is divisible, Is determined using the character information as to whether or not each character can be combined with an adjacent character, and if combined, creates a character string candidate using the characters before and after combining, respectively. Means, analyzing means for analyzing each of the character string candidates created by the combining means using the word information, and extracting means for extracting an appropriate character string as a result of the analysis by the analyzing means. An analyzer characterized by the above-mentioned. 10. An analyzing apparatus for analyzing a character string obtained by character recognition of a character image read by an image reading means, comprising: data indicating whether each character can be divided / combined for a plurality of characters; A first storage unit for storing character information including data indicating candidates after division / combination when division / combination is possible; and a second storage unit for storing word information representing data relating to a plurality of words.
Storage means, and, as a result of character recognition, determine whether each character obtained is divisible by using the character information, and if divisible, divide the character; before and after division A dividing unit that creates a character string candidate using each character; an analyzing unit that analyzes each of the character string candidates created by the dividing unit using the word information; An extracting device for extracting a simple character string. 11. An analyzing apparatus for analyzing a character string obtained by character recognition of a character image read by an image reading means, comprising: data indicating whether each character can be divided / combined for a plurality of characters; A first storage unit for storing character information including data indicating candidates after division / combination when division / combination is possible; and a second storage unit for storing word information representing data relating to a plurality of words.
Storage means, as a result of character recognition, determine whether each obtained character can be combined with an adjacent character using the character information,
When combining is possible, combining means for creating character string candidates using characters before and after combining, respectively, and analyzing each character string candidate created by the combining means using the word information An analyzing apparatus comprising: an analyzing unit; and an extracting unit configured to extract an appropriate character string as a result of the analysis by the analyzing unit. 12. The character information includes data of each character in each of a left-right direction and a vertical direction. 11. The analyzing apparatus according to claim 9, wherein when the direction of a character string obtained by character recognition is vertical, the vertical character information is used. 13. The character information includes data of each character in each of a left-right direction and a vertical direction, and the combining unit outputs a right-left direction when a character string obtained by character recognition is a left-right direction The analyzer according to claim 9, wherein the character information in the up-down direction is used when the direction of the character string obtained by character recognition is the up-down direction using the character information. 14. The word information includes data relating to the word of the address and place name storing the notation of the address and place name, and further includes means for setting the data relating to the word of the address and place name by the analyzing means to be preferentially used. An analyzer according to any one of claims 9 to 13, wherein: 15. The analyzing apparatus according to claim 9, further comprising a display unit for displaying a character string extracted by said extracting unit. 16. A program executable by an information processing apparatus having a program code for realizing the analysis method according to claim 1. Description: 17. A program executable by a computer device, wherein the computer device executing the program causes the computer device to function as the analysis device according to any one of claims 9 to 15. 18. A storage medium storing the program according to claim 16. Description: 19. A data used in the analysis method according to any one of claims 1 to 8, which indicates whether each character can be divided / combined, and a case where division / combination is possible. A storage medium for storing character information including data indicating candidates after division and combination.