KR20100103351A - Character recognition apparatus, character recognition program and character recognition method - Google Patents

Abstract

PURPOSE: A character recognition apparatus, a character recognition program and a character recognition method are provided to recognize image data including a character at a high speed. CONSTITUTION: A ROM(12) stores matching data for text recognition, and an image input unit(3) acquires image data. A CPU(11) recognizes the character within the image data based on the matching data. The matching data includes first and second recognition matching data which are discriminated from each other. The second recognition matching data relate to a second recognition character which has more characters than the first recognition characters.

Description

CHARACTER RECOGNITION APPARATUS, CHARACTER RECOGNITION PROGRAM AND CHARACTER RECOGNITION METHOD}

The present invention relates to, for example, a character recognition apparatus, a character recognition program, and a character recognition method for recognizing characters contained in image data.

The number of characters in Japanese and Chinese is much higher than in English, and there are more than 3,000 characters of the first level kanji and more than 6000 characters of the second level kanji, respectively. For this reason, when it is going to recognize characters, such as Japanese and Chinese, the process of matching each character with more than 3000-6Q00 in the recognition process of each character is needed. As a result, heavy processing that puts a load on the CPU becomes a serious problem, and the prolonged processing time becomes a serious problem, especially in a poor CPU environment such as a portable device.

On the other hand, as a character recognition object, there exists a character printing medium in which description content is determined to some extent, such as a business card. As an apparatus for reading the character of such a character printing medium, the apparatus which recognizes the character of a business card is proposed (refer patent documents 1-3).

The business card recognition device of Patent Literature 1 designates a range of recognized images to a user with a pen, and designates to which attribute (name, company name, address, etc.) the recognized characters of this range are registered.

The character recognition device of Patent Document 2 registers a plurality of areas requiring character recognition as a template, selects this template, and then recognizes the characters in the area determined by the template.

The character recognition apparatus of patent document 3 designates a recognition target area | region by a keyboard, and recognizes a character with respect to this target area | region.

Neither apparatus processes the entire business card at once, but only the categories designated by the area. For this reason, the recognition process can be promptly performed about the category desired by a user. That is, since the range for performing the recognition process is narrowed down to a category desired by the user, the recognition process can be speeded up and the load on the CPU can be suppressed.

However, such a prior art shortens the time required for the recognition process by narrowing the character recognition range, but does not shorten the time for the character recognition process itself in the narrow range.

In addition, the above-described prior art has a problem that it takes time and is troublesome. In detail, the prior art needs to specify a range (or template) for character recognition for each business card, and set which category the characters in the range correspond to. For this reason, the larger the number of business cards you want to input, the more complicated the work is required from the user.

On the other hand, a business card which accepts a business card as image data by the terminal device, transmits the received image data to the server via the communication network, performs character recognition at the server, and transmits the recognition result from the server to the terminal device again via the communication network. 0CR treatment system is proposed (refer patent document 4).

According to this method, even if the CPU of the terminal device is poor, the time required for the recognition processing can be shortened if the CPU of the server is high in processing capacity.

However, the method using such a communication network takes communication time. For this reason, there is a problem that the communication time becomes longer depending on the degree of congestion of the line. In addition, since communication becomes essential, a problem arises that, for example, a portable terminal that performs wireless communication such as a mobile telephone cannot be used in a place where radio waves do not reach, such as underground.

Japanese Patent Application Laid-Open No. 10-55413 Japanese Patent Laid-Open No. 2001-202475 Japanese Patent Laid-Open No. 2-240787 Japanese Patent Laid-Open No. 2003-296353

SUMMARY OF THE INVENTION In view of the above-described problems, the present invention provides a character recognition device, a character recognition program, and a character recognition method for realizing high-speed character recognition with a simple operation on image data containing characters, and satisfying the user. It aims to improve.

The present invention provides a character recognition comprising storage means for storing matching data for character recognition, image data acquisition means for acquiring image data, and character recognition means for recognizing characters in the image data based on the matching data. The apparatus for matching the second data includes: first matching matching data for a first predetermined character and a second recognition character including characters other than the first recognition character and having a larger number of characters than the first recognition character. The second recognition matching data for a character is configured to be distinguishable, and the character recognition means includes a first character recognition process for performing character recognition on the image data by the first recognition matching data, 1 Recognition necessity determination process for determining whether the result of the character recognition process corresponds to the recognition condition, and the corresponding recognition required And a character recognition device configured to execute a second character recognition process for performing character recognition with the second recognition matching data, for a recognizing object determined that re-recognition is required by the presence or absence determination process.

The matching data can be composed of data used for matching of character recognition, such as configuration of feature data representing the characteristics of characters.

The storage means can be constituted by means for storing data such as a hard disk or a nonvolatile memory.

The image data may be data of an appropriate image such as a color image, a gray scale image, or a monoclonal two-tone image.

The image data acquisition means may be, for example, an imaging device such as a CCD camera or a CMOS camera, or a USB (Universal Serial Bus) or LAN that receives data of a character string image captured by such an imaging device from the imaging device. It can be configured by appropriate means capable of acquiring image data, such as a communication interface such as a (Local Area Network).

The character recognition means extracts feature data from, for example, a character image in the image data, and applies the feature. The data may be matched with the feature data of the matching data and used as a means for recognizing which character.

The first recognition character can be, for example, an appropriate character determined in advance, such as an alphanumeric character, a category determination character, or a plurality thereof.

The said 2nd recognition character can be made into the appropriate appropriate character previously determined, for example, hiragana, katakana, kanji, or plural of these.

The character recognition device is constituted by an apparatus such as a semiconductor chip having an information processing unit such as a CPU for executing information processing and an information storage unit such as a memory for storing information and an interface unit such as a connection terminal for connecting to an external device, or And an information processing terminal such as a cellular phone, PDA (Porsonal Digital Assistants), a personal computer, or the like having an imaging means, a control means and a storage means.

According to the present invention, it is possible to realize high-speed character recognition by simple operation on image data, and to improve user's satisfaction.

As an aspect of the present invention, the character recognition means is configured to perform a category discrimination process for discriminating a category of a character group in the image data, wherein the first recognition matching data is for category discrimination used for the category discrimination. It may be a configuration that includes characters.

The category determination character is,. For example, it can be comprised by the character used for category determination, such as E-Mai1 "," URL "," TEL "," FAX "," Postal code ", or the character used for the plurality of these.

This aspect makes it possible to also perform category discrimination in the first character recognition process.

In addition, as an aspect of the present invention, the first recognition matching data may be configured to include a part letter obtained by dividing the corresponding combination by elements for a combination character divided into two or more parts among the category determination characters.

The combined character is a character consisting of one piece and a room when the direction of the character string is a left and right direction, or a character consisting of a tube and an angle when the character direction is an up and down direction. can do. Specifically, it can be set as "話", "住", "所", and the like.

The part character is a character corresponding to a side or a room of the combined character when the character string direction is in the left and right direction, or a tube of the combined character when the character string direction is in the up and down direction, or It can be set as a letter consisting of squares. Specifically, for example, "言" and "舌", "人", "Master", "Ho", "Geun", etc. can do.

Thereby, even when the combined character of one character is mistaken as a two-letter part character, the category can be correctly determined in the misrecognized state. In detail, for example, a single combination character "話" is recognized as two component letters "言" and "舌", and the character string is "電 言 舌". Can be recognized as a category of "telephone".

In addition, as an aspect of the present invention, the character recognition means determines whether or not a mistaken character that should not be used in the category is included in the character group in which the category is determined in the category determination process, and is included. In this case, a correction process for correcting the mistaken character with the character used in the category can be performed.

Thereby, for example, the character recognized by the letter "Z" in the category of "telephone number" can be corrected by "2" of the number, and the recognition accuracy can be improved efficiently.

In addition, as an aspect of the present invention, the correction processing can be configured to correct by again performing character recognition, targeting only the characters used in the category when the mistaken characters are included.

As a result, character recognition can be performed using only the data of the character used in the category among the matching data, so that character recognition can be performed at high speed and with high accuracy.

Moreover, as an aspect of this invention, the said 1st character recognition process is a structure which employ | adopts the letter with the highest matching reliability by obtaining a several character candidate from the thing with high matching reliability, The said correction process contains the said mistaken character When there is a character, the character candidate of the mistaken character is corrected to the character having the highest matching reliability among the characters used in the category.

Thereby, the correct character can be corrected without executing character recognition again by the matching data, and character recognition can be completed at high speed and with high accuracy.

The present invention further provides a second recognition character including first recognition matching data relating to a predetermined first recognition character and a character other than the first recognition character and having a larger number of characters than the first recognition character. Image data acquiring step of acquiring image data by means of image data acquiring means, and storing matching data configured to discriminate second matching data for recognition in relation to the storage means; A first character recognition step of performing character recognition in the character recognition means with respect to the image data, a re-recognition necessity determination step of determining in the character recognition means whether or not the result of the first character recognition processing corresponds to a recognition condition; The second recognition on the target of the recognizing which has been determined to require the recognizing by the re-recognition necessity determination processing; The character recognition program which causes a computer to execute a 2nd character recognition step which performs character recognition by said literacy recognition means by the dragon matching data is possible.

The character recognition program can be stored in a suitable storage medium such as a CD (Commpact Disc), a DVD (Digital Versatile Disk), a flexible disk, a memory, a chip, or a hard disk.

According to the present invention, when a character recognition program is installed in a computer, it is possible to realize high-speed long-term character recognition with a simple operation on image data.

The present invention also relates to a first recognition matching data relating to a predetermined first recognition character and a second recognition character that includes characters other than the first recognition character and has a larger number of characters than the first recognition character. Image data acquiring step of acquiring image data by means of image data acquiring means, storing matching data configured to distinguish the second recognition matching data relating to the image recognition means, and the image using the first recognition matching data. A first character recognition step of performing character recognition in the character fusion means on the data, a re-recognition necessity determination step of determining in the character recognition means whether or not the result of the first character recognition process corresponds to a recognition condition; The second recognizing solvent for the recognizing target determined by the recognizing whether or not the recognizing is required by the recognizing process. By the data may be a character recognition method for executing the second character recognition step of executing the character recognition by the character recognition means.

Thereby, high speed character recognition can be performed with respect to image data by simple operation.

According to the present invention, it is possible to provide a character recognition device, a character recognition program, and a character recognition method for realizing high-speed character recognition with a simple operation on image data such as a photographed business card image, thereby improving user satisfaction. .

1 is a perspective view of a mobile terminal.
2 is a block diagram showing a configuration of a mobile terminal.
3 is an explanatory diagram for explaining a configuration of dictionary data;
4 is a flowchart of character recognition processing by a character recognition program.
5 is an explanatory diagram of a photographed image and a recognized character for character recognition;

One embodiment of this invention is described with drawing below.

[Example]

FIG. 1: shows the perspective view of the portable terminal 1, FIG. 2 shows the block diagram of the structure of the portable terminal 1, and FIG. 3 shows the block diagram of prior data.

As shown in FIG. 1, the portable terminal 1 is provided with an image input unit 3, an operation unit 5, and a screen display unit 7. The image input unit 3 is a digital camera for capturing an image, the operation unit 5 is a button to be pressed, and the screen display unit 7 is a liquid crystal display or an organic EL display that displays an image.

As shown in FIG. 2, the portable terminal 1 includes, in addition to the image input unit 3, the operation unit 5, and the screen display unit 7, a CPU 11, a R0M 12, and a RAM 13. This is connected to the bus 16 and provided.

The image input unit 3 is a digital camera such as a CCD camera or a CMOS camera and is connected to the bus 16. The picked-up image obtained by picking up the image input part 3 is transmitted to CPU11 as digital image data. This digital image data may be finally stored in a separate storage device (nonvolatile memory or the like). In addition, the image input unit 3 receives an imaging signal from the CPU 11 and starts imaging.

The operation unit 5 is connected to the bus 16. This operation part 5 transmits the pressure reduction signal operated by the user to the CPU 11 via the bus 16.

The screen display unit 7 has a function as an operation screen display unit 7a for displaying an operation screen and a function as a recognition result display unit 7b for displaying character recognition recognition results. Is connected to the bus 16. When image data is sent from the CPU 11, an image of this image data is displayed on the operation screen display unit 7a or the recognition result display unit 7b.

The CPU 11 is a processor that performs various operations by using the RAM 13 as a temporary storage area according to a program stored in the R0M 12, a control function 21 for controlling each part, and a character recognition function ( 22)

The character recognition function 22 includes a character string extraction processor 31, a character extraction processor 32, a character recognition processor 33, a recognition reliability determination processor 34, a category determination processor 35, and a recognition result correction processor. Has 36.

The character string extraction processing unit 31 performs a process of extracting a character string from the image data. Extraction of the character string is performed by an appropriate method, for example, by binarizing image data to obtain projective data, and distinguishing the character string from the margin from the projective data. Specifically, for example, a histogram in the vertical direction is calculated from a black pixel number for each pixel row for horizontal writing characters such as a business card and the like, and a range in which a predetermined number or more of black pixel rows are contiguous is used as a character string, or A histogram in the left and right directions is calculated from the black pixel number of each pixel column for the vertical writing characters such as the recipients of the postcard, and the execution is performed by an appropriate method such as setting the range in which the columns of the predetermined number or more of black pixels are continuous.

The character extraction processing part 32 performs the process of extracting a character from the said string. This character extraction calculates a histogram in the left and right directions from the black pixel number of each pixel column with respect to the horizontal writing character, and sets the range in which the predetermined number or more of the black pixels are contiguous to one character or the pixel row for the vertical writing character. A histogram in the vertical direction is calculated from the number of black pixels of a star, and the execution is performed by an appropriate method, such as setting a range in which a predetermined number of black pixels are continuous to one character.

The character recognition processing unit 33 executes a process of recognizing a character and a process of obtaining matching reliability by determining which character matches or approximates matching information using matching information of the matching dictionary data 24. The processing for recognizing these characters includes a specific character recognition process for recognizing only a specific character centered on an alphanumeric character at high speed, and an all-character recognition process for recognizing all characters.

The recognition reliability determination processing unit 34 executes a process of determining whether or not the matching reliability (recognition reliability) obtained by the character recognition processing unit 33 exceeds a predetermined threshold.

The category determination processing unit 35 executes a process of determining which category the character string belongs to using the category determination dictionary data 25.

If there is a character in the character string that is not used in the category, the recognition result correction processing unit 36 executes a process of correcting this character to the character used in the category.

The R0M 12 stores the character recognition program 23, the matching dictionary data 24, and the category determination dictionary data 25.

As shown in Fig. 3A, the matching dictionary data 24 is composed of a character code, matching information, and a target category.

The character code is, for example, a character code of an appropriate method such as Shift-JIS.

The matching information stores character data of characters for each character.

A plurality of types of target categories are registered for one character. Specifically, "E-Mai1", "URL", "TEL", "FAX", "postal code", etc. are registered as a target category.

The data stored in the matching dictionary data 24 can be distinguished by the first recognition alphanumeric character portion 24a, the first recognition category character portion 24b, and the second recognition character portion 24c. Is remembered.

The first recognition alphanumeric character part 24a mainly consists of Western characters such as English, numbers, and symbols.

The first recognition category character part 24b is a character used for identification of a category, that is, "E-Mai1", "URL", "TEL", "registered as a target category of the matching dictionary data 24". It consists of characters used in either of "FAX" and "Zip code".

Specifically, "メ (me)", "-", "ル (ru)", "ホ (ho)", "mu (no)", and "ぺ (pe) which become a key code indicating that it is a target category. ``, `` Ji '', `` 電 (pre) '', `` (Tue) '', `` (f) '', `` (a) '', `` (chu) '', `` (ku) '' , Characters such as "s", "郵", "便", "番" and "번" are included. In addition, regarding the combined characters such as `` Japanese '' and `` Japanese '' among these characters, `` language '' using a side and a room of the part as one character , Parts characters such as 설, 人, 갱, and gang are also included.

In addition, it is preferable to make the part letter which isolate | separated the combined letter into the combined letter by which the part letter is arranged at least in the direction of the character string to be recognized. This character string direction can be made into the appropriate direction in which the characters are arranged, such as the longitudinal direction if the characters are arranged in the longitudinal direction, and the transverse direction if the characters are arranged in the horizontal direction.

The 2nd recognition character part 24c is comprised from the character (mainly Japanese character) which does not enter in either of the 1st recognition alphanumeric part 24a and the 1st recognition category character part 24b.

In addition, although this 2nd recognition character part 24c is set as the structure which does not contain the 1st recognition category character part 24b, a part of 2nd recognition character part 24c is used for a 1st planting category It is good also as a structure made into the character part 24b.

As shown in Fig. 3B, the category determination dictionary data 25 is composed of a plurality of files for each category, and a character string for discriminating as a category is stored in each file. In the character string, a character string containing a combination character as a part character, such as "telephone" and "electron word", is also stored. The character string stored in the category determination dictionary data 25 may be a sequence of character codes.

The RAM 13 temporarily stores data used by the CPU 11 for various controls and calculations, and stores images such as sentence string images in the image memory 14.

The portable terminal 1 is also provided with a communication antenna for wireless communication, a speaker for audio output, a microphone for sound collection, and the like. Thereby, it is comprised so that it can be used as a portable telephone.

In addition, as shown by a virtual line in FIG. 2, CPU11 and R0M12 may be mounted in one chip 15, and may be comprised as an electronic component. In this case, if the chip 15 is provided in the portable terminal 1, the character recognition function can be added easily.

FIG. 4 is a flowchart of character recognition processing executed by the CPU 11 of the portable terminal 1 in accordance with the character recognition program 23. FIG. 5 is an explanatory diagram of a photographed image and a recognized character that have been character recognized.

First, as a preliminary preparation, when the character recognition processing such as a business card reading mode is selected in response to a user's operation input, the CPU 11 makes a state in which the image can be photographed by the image input unit 3, and the screen display unit 7 The operation screen display unit 7a shown in Fig. 5A is displayed.

The operation screen display unit 7a includes a recognition screen display unit 71 indicating that the screen is a character recognition screen, and a preview image display unit 72 for displaying a preview image captured by the image input unit 3 (camera). ), A return button 73 for returning to the previous screen, a recognition button 74 for starting the recognition process, and a submenu button 75 for displaying a submenu.

When the recognition button 74 is selected by the user, the CPU 11 executes image capturing by the image input unit 3 to obtain a character image (step S1).

The CPU 11 executes a character string extraction process by the character string extraction processing unit 31 to extract a character string from the character image (step 52). At this time, a plurality of character strings such as a character string of the company name display portion, a character string of the name display portion, and a character string of the address display portion are obtained.

The CPU 11 initializes by assigning "0" to the variable "i" (step S3), and executes the first character recognition process by the character recognition processing unit 33 on the i-th extracted character string (step) S4). In this first character recognition processing, the first recognition alphanumeric character part 24a and the first recognition category character part of the matching dictionary data 24 are used for each character extracted from the character string by the character extraction processing part 32. Character recognition is performed using only 24b.

In addition, the character extraction by the character extraction processing part 32 is corresponded to the process which extracts the boundary of a character and a character. For example, with respect to a character string image, the character portion is binarized so that the character portion is black, the black pixel is projected in the direction perpendicular to the character string, and the cutout of the projected data is used as a candidate for the boundary between the character and the character.

Since the first recognition alphanumeric character portion 24a is composed mainly of alphanumeric characters, and the first recognition category character portion 24b is composed mainly of characters used for category determination, a matching dictionary The amount is very small compared to the total number of characters of the data 24. For this reason, the first character recognition process is completed at a much higher speed than the second character recognition process described later.

The CPU 11 determines the matching reliability in the first recognition processing by the recognition reliability determination processing unit 34 (step S5). Here, the determination of the matching reliability checks, for example, the matching reliability in character units, and when the matching reliability is lower than a predetermined threshold as a recognition condition with respect to the character with the lowest matching reliability, the matching reliability of the character string is 0K, Otherwise, it can be NG. In addition, the average value of the matching reliability of each character in the character string is calculated, and if the average value is equal to or greater than a threshold value as a predetermined recognition condition, the matching reliability of the character string may be 0K, otherwise, NG may be set.

In determining the matching reliability, a predetermined number of character candidates may be stored for each character in the order of high matching reliability.

If the determination of the matching reliability is 0K (step S6: Yes), the CPU 11 proceeds to step S8.

If the determination of the matching reliability is NG (step S6: No), the CPU 11 executes the second character recognition process by the character recognition processing unit 33 on the i-th extracted character string (step S7). In this second character recognition process, character recognition is performed using all the data of the matching dictionary data 24.

It is also possible to perform character recognition only by the second recognition character portion 24c, and to configure a character that recognizes the character with the highest matching reliability including the matching reliability in the first character recognition process.

In this second character recognition process, when extracting a character from a character string, the boundary candidates of the characters are examined so that the combined character is not misrecognized as a plurality of part characters, and one character is cut out more accurately than the first character recognition process. May be configured. In this case, the character length of one character may be recognized, and the boundary may be recognized in units of the corresponding character length. This character length indicates the length of one character in the character string direction and can be said to be, for example, the number of pixels in the character string direction in one character. Thereby, the recognition precision of a character can be improved further.

The CPU 11 adds 1 to the variable "i" (step S8), and repeats the process to step S4 until "i" reaches the number of extracted character strings (step S9: Yes).

When "i" reaches the number of extracted strings (step S9: N0), the CPU 11 initializes by assigning "0" to the variable "i" (step S10), and the category discrimination processing unit about the i-th extracted string. (35), category determination processing is executed (step S11).

This category determination processing is performed by determining whether the sequence of characters recognized in steps S4 and S7 matches the sequence of characters of each category stored in the category determination dictionary data 25. If they match (when matching reliability is higher than a predetermined threshold), it is determined as that category.

The CPU 11 performs the recognition result correction processing by the recognition result correction processing unit 36 (step S12). This recognition result correction process determines whether the i-th extraction character string does not contain the characters not used in the category determined in step S11, and corrects if it contains.

The recognition result correction processing unit 36 that performs this correction corrects the characters used in the category with the characters having the highest matching reliability, based on the matching new amount of the plurality of character candidates stored in step S5. If there is no character with high matching reliability in the character used in the category, character recognition processing is performed again for only the character used in the category, and the character is corrected.

In addition, this correction | amendment may be performed by a suitable method, such as performing character recognition process again using only the character of the category among the matching dictionary data 24. FIG.

The CPU 11 adds 1 to the variable "i" (step S13), and repeats the process to step S11 until "i" reaches the number of extracted character strings (step S14: Yes). When "i" reaches the number of extracted character strings (step S14: NO), the CPU 11 displays the recognition result display unit 7b shown in Fig. 5B on the screen display unit 7 and ends the processing.

By this character recognition process, if it is the example shown in FIG. 5, it processes as follows.

First, each character string of (1)-(9) is recognized by character string extraction process (step S2), as shown to FIG. 5 (B).

By executing the first character recognition processing (step S4) in this order for each character string of (1) to (9), the character is recognized in the first character recognition character portion 24a and the first character category character portion 24b. Character recognition of recognizable categories can be performed. That is, the postal code of (4), the telephone number of (6), the FAX of (7), the E-mai1 of (8), and the URL of (9) can be recognized.

At this time, since the part letters obtained by separating the combined characters into parts are also registered in the first character category 24b for recognition, as shown in Fig. 5C, the definite characters "話 (hwa)" Even when characters are separated by "言" and "舌", they are recognized as "言" and "舌".

Since (1) (2) (3) and (5) of FIG. 5B have low matching reliability, the second character recognition process (step S7) is executed to perform character recognition using all characters. Thereby, the character recognition can be performed also with respect to (1) (2) (3) (5) of FIG.

Thereafter, the category of each character string can be determined by the category discrimination process (step S11). At the time of this category discrimination, the category in which the combined character is included, such as "電話 (telephone)", is also a character string containing the combined character as a part character, as in "電 言 舌". Since it is registered in, the correct category can be easily recognized with high accuracy.

That is, for example, if the combined character is to be correctly recognized, a process of examining the boundary candidates of the characters is required in order to avoid making them into two characters when extracting the characters, resulting in a large amount of calculation. However, even if such processing is not carried out, since two letters are recognized as they are and categories are determined, the boundary of the letters is roughly determined, and the processing can be performed at such a high speed. Since the time reduction amount of this boundary process is much larger than the time increase amount by the increase of the comparison target for recognizing the combined characters as a collection of parts characters, the overall processing time can be shortened.

In addition, the recognition result correction process (step S12) corrects the character string of the determined category by the recognition excluding the character not used in the category, so that character recognition with high precision can be performed.

Specifically, for example, the "telephone (telephone)" of (6) of FIG. (075) 1234-5678 "is a character string whose category is" TEL. " When this is misidentified as "電話: (075) 1Z34-5678" in the first character recognition processing, since "Z" is a character that is not used in the category "TEL", correction processing is performed on this. Correction is made to "2" used in the category "TEL". As a result, the accuracy of the recognition can be improved, and the output of the recognition result in which the user feels discomfort can be suppressed.

By the above configuration and operation, it is possible to realize high-speed character recognition by simple operation on image data, and to improve user's satisfaction.

After executing the first character recognition process with limited characters, the second character recognition process is executed only for the character strings whose matching reliability is lower than the threshold value, so that the load of arithmetic processing can be reduced and a language having a large number of characters such as Japanese or Chinese. Even a character can be recognized at high speed as a whole. In particular, even a poor CPU can recognize characters sufficiently fast.

In detail, in the first character recognition processing, as the first recognition alphanumeric character portion 24a, only Western characters (about 100 characters) such as English, numbers, and symbols, and fewer first character recognition characters Since only (24b) is used, the processing can be performed several times or more faster than the case where the matching is performed on all characters.

In addition, in the reading of a business card, a character string indicating "E-Mail", "URL", "TEL", "FAX", and "Postal code" is composed of characters of Western and American systems such as rough English, beef, and symbols. . Among them, in particular, "E-Mail" and "URL" are often strings compared with character strings representing other things such as "full name" and "company name". Therefore, among the business cards, it is common that a large percentage of Western and American characters such as English, numbers, and symbols occupy. The first E-Mail, URL, and TEL composed of these characters are executed by first performing a lightweight first character recognition process targeting only Western characters such as English, numbers, and symbols with a large ratio. , The character strings indicating "FAX" and "zip code" can be determined. As for the other character strings only, by performing the heavy second character recognition processing targeting Japanese, Chinese, etc., as a result, the processing speed of the card reading as a whole can be improved.

In the first character recognition process, the first character category character part 24b also targets the characters of the keyword representing the category. Therefore, "E-Mail", "URL", "TEL",! FAX "," The zip code "can be reliably recognized in the first character recognition process.

In detail, as a keyword indicating a category such as "E-Mai1", "URL", "TEL", "FAX", "Postal code", etc., "メ (Me)", "-", "ル" "," ホ (ho) ",

The card may contain characters other than the Western system such as "mu", "페", and "ジ". For example, "Me- ル (e-mail): abc @ ****, co.jp", "ホ-ム ぺ-ジ (homepage): http; // www, ****, co.jp" And the like.

For this reason, if there is no character of the 1st recognition category character part 24b at the time of a 1st character recognition process, the matching reliability at the time of recognition will worsen, and "E-Mai1", "URL", "TEL", The case where the character strings indicating "FAX" and "zip code" cannot be determined at first. On the other hand, this problem can be solved by using the 1st category character part 24b for recognition.

In addition, the necessity of examining the boundary of characters can be reduced, and the processing can be speeded up. In detail, when it tries to recognize correctly the combined characters, such as "hwa", which consists of a piece and a room, it is generally a part character "language" divided into individual parts. The process of examining a boundary is necessary so that it may not be recognized as two characters of "tongue". However, in the case of Western and American characters such as English, beef, and symbols, there is no such combined character, so even though it is not necessary to examine the boundary, "E-Mai1", "URL", "TEL", "FAX" In order to recognize characters in the category of "postal code", the processing time becomes longer if the boundary examination processing is performed on all characters for the combined characters such as "telephone" which are rarely mixed.

On the other hand, in the first character recognition process, since the part character obtained by separating the combined character into parts is also registered, it is not necessary to examine the boundary of the character, and the processing can be performed at high speed.

In particular, the inclusion of the combined characters is limited to display parts indicating categories such as the display of "telephone" or "zip code", for example, and information of the contents of the category (actual phone number or postal code, etc.). ) Is not included. The display portion of this category may be recognized as being in the "telephone" category even if it is recognized as "electronic speech". By using this property, the part text that separates the combined text is also registered and recognized. Therefore, without causing the harm caused by recognizing the combined text as the part text (for example, `` Del. Even if it recognizes it, it does not register the letter "電 言 舌", and it can process it at high speed.

In addition, since the male character recognition can be performed with respect to the character string which recognized the category, the recognition accuracy can be improved, and the output of the recognition result in which the user feels discomfort can be suppressed.

In addition, since the character string extraction processing unit 31 extracts the character string and determines the category from the category determination processing unit 35 to the category, the user designates the character recognition area by operation or selects and operates the category of the specified area. The labor can be reduced and the convenience can be improved.

In addition, the character recognition program 23 is stored in a storage means of a server connected to a telecommunication line such as the Internet, and can be downcoded via a telecommunication line or an ASP (App1imtim Service Provider) via a telecommunication line. It is also possible to set it as the structure which can be used as a service. In this case, although a communication load is applied, the load of the arithmetic processing of a server can be reduced.

In the configuration of the present invention and the correspondence of the above-described embodiment,

The character recognition apparatus and the computer of the present invention correspond to the mobile terminal 1 of the embodiment.

Likewise below,

The image data acquisition means corresponds to the image input unit 3,

The character recognition means corresponds to the CPU 11,

The storage means corresponds to R0M 12,

The matching data corresponds to the matching dictionary data 24,

The first recognition matching data corresponds to the first recognition alphanumeric character portion 24a and the first recognition category character portion 24b,

The second recognition matching data corresponds to the first recognition alphanumeric character portion 24a, the first recognition category character portion 24b, and the second recognition character portion 24c,

The character for category determination corresponds to the first category character portion 24b for recognition,

The image data taking step corresponds to step S1,

The first character recognition processing and the first character recognition step correspond to step S4,

The recognition necessity determination process and the recognition necessity determination step correspond to steps S5 to S6,

The second character recognition processing and the second character recognition step correspond to step S7,

The category determination process corresponds to step S11,

The correction process corresponds to step S12,

The first character for recognition corresponds to English, numbers, symbols, characters for categories,

The second person character supports the expert,

This invention is not limited only to the structure of embodiment mentioned above, Many embodiment can be obtained.

INDUSTRIAL APPLICABILITY The present invention can be used for an apparatus for recognizing a character, and in particular, for an apparatus for recognizing a character from a character display medium in which description items (categories) are determined to some extent, such as reading a business card, reading a recipient name of a postcard or an envelope. Can be.

1: mobile terminal
3: image input unit
11: CPU
13: RAM
2 yo: character recognition program
24: Dictionary data for matching
24a: first alphanumeric character part for recognition
24b: Category character part for first recognition
24c: second character for recognition

Claims (8)

Storage means for storing matching data for character recognition;
Image data acquiring means for acquiring image data;
A character recognition apparatus comprising character recognition means for recognizing a character in said image data based on said matching data,
The matching data,
First matching data for recognition relating to a first predetermined character for recognition,
A second recognition matching data relating to a second recognition character including characters other than the first recognition character and having a larger number of characters than the first recognition character, is configured to be distinguishable
The character recognition means,
First character recognition processing for performing character recognition on the image data by the first recognition matching data;
Re-recognition necessity determination processing for determining whether the result of the first character recognition processing corresponds to a re-recognition condition, and
A character characterized in that it is a structure which performs the recognition process which is the 2nd character which performs character recognition with the said 2nd recognition matching data with respect to the recognition object determined by the said recognition necessity determination process by the said recognition necessity. Recognition device. The method of claim 1,
The character recognition means is configured to perform category determination processing for determining a category of a character group in the image data,
And the first recognition matching data is configured to include a category determination character used for the category determination. The method of claim 2,
The first recognition matching data is a character recognition device comprising a part letter obtained by dividing a corresponding part by element for a combined character divided into two or more parts among the category determination characters. 4. The method according to claim 2 or 3,
The character recognition means,
A correction process of determining whether or not a mistake character is to be used in the category in the character group in which the category is determined in the category determination process, and correcting the misinterpretation character to the character used in the category if it is included. Character recognition apparatus characterized in that the configuration for executing. The method of claim 4, wherein
The correction process,
And a character that is corrected by performing character recognition again, targeting only the characters used in the category when the mistaken character is included. The method of claim 4, wherein
The first character recognition process is a configuration in which a plurality of character candidates are obtained from a high matching reliability, and a character having the highest matching reliability is adopted.
The correction process,
And when the mistake character is included, correcting the character with the highest matching reliability among characters used in the category among character candidates of the mistake character. First recognition matching data relating to a first predetermined character for recognition and second recognition for a second character for recognition including characters other than the first character for recognition and having more characters than the first character for recognition. Matching data configured to distinguish the matching data is stored in the storage means;
An image data acquisition step of acquiring recall data by the image data acquisition means,
A first character recognition step of performing character recognition in character recognition means on said image data by said first recognition matching data;
A re-recognition necessity determining step of judging by the character recognizing means whether or not the result of the first character recognition process corresponds to a re-recognition condition;
Causing the computer to execute a second character recognition step of performing character recognition in the character recognition means by the second recognition matching data on the re-formation target determined by the re-recognition necessity determination process; Character recognition program, characterized in that. First recognition matching data relating to a first predetermined character for recognition and second recognition for a second character for recognition including characters other than the first character for recognition and having more characters than the first character for recognition. Matching data configured to distinguish the matching data is stored in the storage means;
An image data acquisition step of acquiring image data by the image data acquisition means,
A first character recognition step of performing character recognition in character recognition means on said image data by said first recognition matching data;
A re-recognition necessity determining step of judging in the character recognition faucet whether or not the result of the first character recognition process corresponds to a re-recognition condition;
And a second character recognition step of executing character recognition in said character recognition means by said second recognition matching data, for the recognizing object that has been determined to require re-recognition by said re-recognition necessity determination process. Character recognition method.

Images