JP2015184691A - Image processor and image processing program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an image processor for extracting a character string even from an image having a background in the case of extracting the character string from in an image.SOLUTION: A plurality of binarization means of the image processor binarize an image, selection means selects a pixel block on the basis of the possibility that the pixel block in each binarized image binarized by the binarization means is a character, extraction means retrieves adjacent characters from the pixel block selected by the selection means to extract a character string, and combination means combines the character string extracted by the extraction means.

Description

  The present invention relates to an image processing apparatus and an image processing program.

  In Patent Document 1, it is an object to provide a character recognition method and apparatus that can be generally applied to character recognition in the field such as outdoors or in a factory, and binarize a captured image of a character to be recognized with a plurality of threshold values. The obtained character pattern is obtained, and the character area is cut out by projecting and adding the plurality of character patterns in the X and Y directions. The obtained character is input to the neural network, and the output value exceeds the set reference value. Is output as the final recognition result.

  In Patent Document 2, it is an object to realize robust character recognition with respect to unknown lighting conditions, and a plurality of different binarization means, means for integrating character pattern candidates obtained from them, and character pattern candidates are analyzed. Are recognized as character strings.

JP 2003-016386 A JP 2006-059124 A

  An object of the present invention is to provide an image processing apparatus and an image processing program that extract a character string from an image with a background when a character string is extracted from the image.

The gist of the present invention for achieving the object lies in the inventions of the following items.
The invention of claim 1 is based on a plurality of binarization means for binarizing an image and a possibility that a pixel block in each binarized image binarized by the binarization means is a character. A selection unit for selecting a pixel block, an extraction unit for searching for adjacent characters from the pixel block selected by the selection unit and extracting a character string, and a synthesis unit for synthesizing the character string extracted by the extraction unit An image processing apparatus comprising:

  The invention according to claim 2 is characterized in that the selection means has a certainty factor of a character as a result of character recognition of each of the binarized images higher or higher than a predetermined first threshold, and the complexity of the character 2. The image processing apparatus according to claim 1, wherein a pixel block whose is higher than or equal to a second threshold value is selected.

  The invention according to claim 3 is characterized in that the complexity is one of the number of alternations in the circumscribed rectangle of the pixel block, the number of strokes of characters as a character recognition result, or a combination thereof. The image processing apparatus according to 2.

  According to a fourth aspect of the present invention, the selection means has a certainty factor of a character as a result of character recognition of each of the binarized images being higher than or equal to a predetermined first threshold, and the character is determined in advance. 2. The image processing apparatus according to claim 1, wherein a pixel block that is not a written character is selected.

  According to a fifth aspect of the present invention, the extraction unit sets a rectangle corresponding to the size of the circumscribed rectangle of the first pixel block selected by the selection unit, and each binarized image around the circumscribed rectangle. Character recognition within the rectangle, and a combination of pixel blocks whose certainty of the result of character recognition is higher or higher than a third threshold is set as a second character next to the first character, and The image processing apparatus according to claim 1, wherein the second character is set as the first character to extract a character string.

  A sixth aspect of the present invention is the image processing apparatus according to the fifth aspect, wherein the extraction means performs the character recognition by excluding a combination of pixel blocks matching a predetermined condition.

  The invention according to claim 7 is, as the predetermined condition, a combination including pixel clusters of different colors, a combination including different stroke widths, and different from a character stroke width in the first pixel blocks. The image processing apparatus according to claim 6, wherein the image processing apparatus is one of a combination including a stroke width, or a combination thereof.

  The invention according to claim 8 is characterized in that the synthesizing unit generates a character string image by performing a logical sum synthesis on the character string images in the respective binarized images, or corresponds to the character string image in each of the binarized images. The image processing apparatus according to claim 1, wherein a character string image is generated on the basis of a result of the counting for each character code of each character.

  The invention of claim 9 is characterized in that the binarization means performs binarization using a plurality of different thresholds, binarization using a plurality of different thresholds after inversion of the image, or binarization of a combination thereof. An image processing apparatus according to any one of claims 1 to 8.

  According to the invention of claim 10, there is a possibility that the computer includes a plurality of binarizing means for binarizing the image, and the pixel block in each binarized image binarized by the binarizing means is a character. A selection unit for selecting a pixel block, an extraction unit for extracting a character string by searching for adjacent characters from the pixel block selected by the selection unit, and a character string extracted by the extraction unit An image processing program for functioning as a synthesizing unit.

  According to the image processing apparatus of the first aspect, when a character string is extracted from an image, the character string can be extracted from an image with a background.

  According to the image processing apparatus of the second aspect, it is possible to select a pixel block which is a character by using the certainty factor of the character as a result of character recognition and the complexity of the character.

  According to the image processing apparatus of claim 3, the number of alternations or the number of strokes can be used as the complexity.

  According to the image processing apparatus of the fourth aspect, it is possible to select a pixel block which is a character by using the certainty factor of the character as a result of character recognition and a predetermined character.

  According to the image processing apparatus of the fifth aspect, it is possible to extract characters around the circumscribed rectangle of the first pixel block.

  According to the image processing device of the sixth aspect, it is possible to perform character recognition by excluding combinations of pixel blocks that match a predetermined condition.

  According to the image processing apparatus of claim 7, as a predetermined condition, a combination including pixel clusters of different colors, a combination including different stroke widths, or a stroke of characters in the first pixel block A combination including a stroke width different from the width can be used.

  According to the image processing apparatus of the eighth aspect, it is possible to generate a character string image based on the logical sum synthesis or the character code count result.

  According to the image processing apparatus of the ninth aspect, binarization can be performed using a plurality of different threshold values.

  According to the image processing program of the tenth aspect, when a character string is extracted from the image, the character string can be extracted from an image with a background.

It is a conceptual module block diagram about the structural example of this Embodiment. It is a flowchart which shows the process example by this Embodiment. It is a conceptual module block diagram about the structural example in the origin character search module of this Embodiment. It is a flowchart which shows the process example by this Embodiment. It is a conceptual module block diagram about the structural example in the character string search module of this Embodiment. It is a flowchart which shows the process example by this Embodiment. It is a conceptual module block diagram about the structural example in the character string search module of this Embodiment. It is a flowchart which shows the process example by this Embodiment. It is explanatory drawing which shows the process example by this Embodiment. It is explanatory drawing which shows the process example by this Embodiment. It is explanatory drawing which shows the process example by this Embodiment. It is explanatory drawing which shows the process example by this Embodiment. It is explanatory drawing which shows the process example by this Embodiment. It is explanatory drawing which shows the process example by this Embodiment. It is explanatory drawing which shows the process example by this Embodiment. It is explanatory drawing which shows the process example by this Embodiment. It is a block diagram which shows the hardware structural example of the computer which implement | achieves this Embodiment.

Hereinafter, an example of a preferred embodiment for realizing the present invention will be described with reference to the drawings.
FIG. 1 shows a conceptual module configuration diagram of a configuration example of the present embodiment.
The module generally refers to components such as software (computer program) and hardware that can be logically separated. Therefore, the module in the present embodiment indicates not only a module in a computer program but also a module in a hardware configuration. Therefore, the present embodiment is a computer program for causing these modules to function (a program for causing a computer to execute each procedure, a program for causing a computer to function as each means, and a function for each computer. This also serves as an explanation of the program and system and method for realizing the above. However, for the sake of explanation, the words “store”, “store”, and equivalents thereof are used. However, when the embodiment is a computer program, these words are stored in a storage device or stored in memory. It is the control to be stored in the device. Modules may correspond to functions one-to-one, but in mounting, one module may be configured by one program, or a plurality of modules may be configured by one program, and conversely, one module May be composed of a plurality of programs. The plurality of modules may be executed by one computer, or one module may be executed by a plurality of computers in a distributed or parallel environment. Note that one module may include other modules. Hereinafter, “connection” is used not only for physical connection but also for logical connection (data exchange, instruction, reference relationship between data, etc.). “Predetermined” means that the process is determined before the target process, and not only before the process according to this embodiment starts but also after the process according to this embodiment starts. In addition, if it is before the target processing, it is used in accordance with the situation / state at that time or with the intention to be decided according to the situation / state up to that point. When there are a plurality of “predetermined values”, they may be different values, or two or more values (of course, including all values) may be the same. In addition, the description having the meaning of “do B when it is A” is used in the meaning of “determine whether or not it is A and do B when it is judged as A”. However, the case where it is not necessary to determine whether or not A is excluded.
In addition, the system or device is configured by connecting a plurality of computers, hardware, devices, and the like by communication means such as a network (including one-to-one correspondence communication connection), etc., and one computer, hardware, device. The case where it implement | achieves by etc. is included. “Apparatus” and “system” are used as synonymous terms. Of course, the “system” does not include a social “mechanism” (social system) that is an artificial arrangement.
In addition, when performing a plurality of processes in each module or in each module, the target information is read from the storage device for each process, and the processing result is written to the storage device after performing the processing. is there. Therefore, description of reading from the storage device before processing and writing to the storage device after processing may be omitted. Here, the storage device may include a hard disk, a RAM (Random Access Memory), an external storage medium, a storage device via a communication line, a register in a CPU (Central Processing Unit), and the like.

  The image processing apparatus 100 according to the present embodiment extracts a character string from an image. As shown in the example of FIG. 1, the image receiving module 110, the binarization (1) module 120A, the binary (2) module 120B,..., Binarization (N) module 120Z, starting character search module 130A, starting character search module 130B,..., Starting character search module 130Z, character string search module 140A, character string .., A character string search module 140Z, and a character string composition module 150.

  The image reception module 110 is connected to a binarization (1) module 120A, a binarization (2) module 120B, a binarization (N) module 120Z, and a character string search module 140Z. The image reception module 110 receives a target image. Accepting an image means, for example, taking an image with a camera or the like, extracting an image from a moving image, reading an image with a scanner or the like, or receiving an image from an external device via a communication line with a fax or the like. And reading out an image stored in a hard disk (including those connected to the computer in addition to those built in the computer). The image is a multivalued image (including a color image). One image may be received or a plurality of images may be received. Further, the contents of the image include a character string such as a design character string in an advertisement pamphlet, a character string described on a signboard in a landscape image, a telop in a moving image, and the like. The character string may have a background. Further, the position, size, color, etc. of the characters are unknown, and it is not necessary to assume that the arrangement of the character strings is a straight line. Specifically, this includes a landscape image taken with a camera, a moving image taken with a lecture in a classroom, an image extracted from a moving image taken with a car drive recorder, etc. (also referred to as a scene image).

There are a plurality of binarization modules 120, starting character search modules 130, and character string search modules 140, respectively. Since each module performs the same process, the binarization (1) module 120A, the starting character search module 130A, and the character string search module 140A will be described as a representative.
The binarization (1) module 120A is connected to the image receiving module 110 and the starting character search module 130A. The binarization (1) module 120A passes the binarized image layer (1) 122A to the starting character search module 130A. The binarization (1) module 120A binarizes the image received by the image reception module 110. Each binarization module 120 performs different binarization processing. As the “different binarization processing”, for example, binarization in which the method itself is different may be used. For example, the binarization (1) module 120A may be a p-tile method, the binarization (2) module 120B may be a mode method, and the binarization (N) module 120Z may be a discriminant analysis method or the like. Any one that outputs different binarized images may be used. Further, as “different binarization processing”, binarization by a plurality of different thresholds, binarization by a plurality of different thresholds after reversing the target image, or binarization of a combination thereof Good. The reason for “inverting the target image” is to deal with white characters and the like.

  The binarized image 900 shown in the example of FIG. 9 is a result of binarization processing performed by any of the binarization modules 120 on a scene image including a signboard on which characters “Nakata store” are written. In the binarized image 900, in addition to the character image “Nakata Shoten”, there are noise (non-character) 910, noise (non-character) 920, and noise (non-character) 930 in the vicinity of the character “ta”. May be included. In the conventional technique, when cutting out a one-character image, it is affected by these noises. As a one-character image, as shown in the example of FIG. A character image may be included including noise (non-character) 910, noise (non-character) 920, and the like.

The starting character search module 130A is connected to the binarization (1) module 120A, the character string search module 140A, the character string search module 140B, and the character string search module 140Z. The starting character search module 130A selects a pixel block based on the possibility that the pixel block in the binarized image layer (1) 122A binarized by the binarization (1) module 120A is a character. The pixel block selected here is a pixel block having a high possibility of being a character. Further, the pixel block includes at least a pixel region that is continuous in four or eight connections, and includes a set of these pixel regions. The set of these pixel areas means that there are a plurality of continuous pixel areas such as 4-connected, and the plurality of pixel areas are in the vicinity. Here, what is in the vicinity includes, for example, those in which the pixel areas are close to each other in distance (the pixel clusters are within a predetermined distance). In many cases, an image of one character is formed as one pixel block. However, it is not necessary that the pixel area is actually recognizable as a character by humans. There are a part of a character, a pixel region that does not form a character, and the like, and any pixel block may be used. The character image selected by the starting character search module 130A becomes the starting character when searching for a character string.
“Select based on the possibility of being a character” refers to selecting a pixel block that has a high possibility of being a character. And "the possibility that it is a character is high" here is that the value indicating the possibility of being a character is equal to or more than a predetermined threshold or the value indicating the possibility of being a character in descending order. Arrangement and higher rank than a predetermined order (for example, the highest value indicating the possibility of being a character). Therefore, if there is no pixel block having a high possibility of being a character in the binarized image, the pixel block is not selected, and the character string cannot be extracted from the binarized image. However, if a pixel block is selected in another binarized image, the character string can be extracted.

The starting character search module 130A determines the certainty of the character as a result of character recognition of the binarized image layer (1) 122A (for example, the distance between the target character image and the character in the dictionary in the feature space for character recognition). May be selected such that the character complexity is higher or higher than the first threshold, and the complexity of the character is higher or higher than the second threshold. Here, the “complexity” is either the number of alternations in the circumscribed rectangle of the pixel block (the number of times of switching from black to white or white to black), the number of strokes of characters that are character recognition results, or these It is good also as a combination.
In addition, the starting character search module 130A has a certainty factor of a character as a result of character recognition of the binarized image layer (1) 122A that is higher or higher than a predetermined first threshold, and the character is determined in advance. A pixel block which is not a written character may be selected.

More specifically, the starting character search module 130A recognizes the connected pixel area as a pixel block by labeling processing already known from each binarized image, and cuts it out as a circumscribed rectangular image. By recognizing the character, the character code of the character candidate and its certainty (accuracy) are obtained. When other pixel blocks are included in the circumscribed rectangle area (when the circumscribed rectangles overlap), the circumscribed rectangle image including the plurality of pixel blocks may be cut out and recognized as characters.
In addition to the character code and the certainty of the character, the number of strokes may be added as an output of character recognition. As the complexity of the character candidate, the number of strokes may be used, or the number of alternating turns of the cut out circumscribed rectangle may be used. The number of alternations may be determined by measuring the number of black and white inversions along several parallel lines and perpendiculars with respect to the circumscribed rectangle, and the maximum number of them is the alternation number.
A predetermined character list may be used instead of the complexity. That is, the characters included in the list are not selected by the starting character search module 130A. Examples of characters included in the list include characters such as “1”, “l”, “|”, “one”, “□”, “L”, “◯”, and the like. The image after the binarization of the scene image includes a non-character image that is not a character but is similar to a character. When the character recognition result is used, the above-mentioned character (“1”) is included in the scene image. Because there are natural objects and structures that can be easily mistaken for, etc., simple-shaped characters are excluded from the starting characters. In addition, characters having these shapes often output a high certainty factor when character recognition is performed, and the starting character cannot be determined only by the certainty factor. There may be a plurality of starting characters.

  The character string search module 140A is connected to the start character search module 130A, the start character search module 130B, the start character search module 130Z, and the character string composition module 150. The character string search module 140A searches for adjacent characters from the pixel block selected by the starting character search module 130 and extracts a character string. A rectangle (hereinafter also referred to as a window) corresponding to the size of the circumscribed rectangle of the first pixel block (hereinafter also referred to as the origin character) selected by the origin character search module 130 is set and is located around the circumscribed rectangle. Character recognition is performed within the rectangle of each binarized image, and a combination of pixel blocks having a certainty level higher than or equal to the third threshold value is recognized next to the first character. Next, the second character is set as the first character, and the character string is extracted. As an outline, a character string is extracted by extracting the surrounding characters of the starting character and tracing the character string one after another using the surrounding characters as the starting character. More specifically, when a window corresponding to the size of the circumscribed rectangle of the starting character is set and the periphery of the circumscribed rectangle is rotated once for all the binarized images, all the pixels that enter the window A method of recognizing a combination of blocks, setting a combination of pixel blocks higher or higher than the third threshold among them as a candidate for the next character, and then setting the next character candidate as the starting character To search for a string.

  Further, the character string search module 140A may perform the character recognition by excluding combinations of pixel blocks that match a predetermined condition. This predetermined condition is a condition for excluding a combination including a different pixel block, and specifically, a combination including a pixel block of a different color or a combination including a different stroke width. A combination including a stroke width different from the stroke width of the character in the first pixel block or a combination thereof may be used. The stroke width here may be a character width.

More specifically, the search for adjacent characters by the character string search module 140A is performed by sliding a window having a size based on the size of the starting character around all the binarized images around the starting character, Character combinations are recognized, and a pixel block combination whose certainty is higher than or higher than the third threshold is set as the adjacent character, and the adjacent character is searched recursively. In this case, combinations of pixel blocks having different colors and combinations having different stroke widths (character widths) are excluded.
Further, combinations in which the width or length of the circumscribed rectangle of the combined pixel block exceeds the range of values determined by the size of the starting character may be excluded.
The adjacent character candidates are limited to the pixel block in the window. Then, assuming that one character is composed of the same color, combinations of different colors may be excluded. A pixel block having a stroke width different from that of the starting character may be excluded because of a high possibility of noise.
In addition, when the width and length of the circumscribed rectangle are large due to the combination of pixel blocks (for example, the width or length of the circumscribed rectangle as a target is compared with the width or length of the circumscribed rectangle of the starting character in advance. If the ratio of the horizontal width to the vertical width of the circumscribed rectangle is larger or larger than a predetermined value), the combination of pixel blocks in the circumscribed rectangle is noise. May be excluded because it is highly likely to contain.
If it can be assumed that the character string is drawn in the same color, the pixel block in the window may be limited to the same color as the starting character.
The character string search module 140 corresponds to an arbitrary character string direction (including a curved line), and also supports a mixed character string.

  The character string synthesis module 150 is connected to the character string search module 140A, the character string search module 140B, and the character string search module 140Z. The character string synthesis module 150 synthesizes the character strings extracted by the character string search modules 140. The character string synthesizing module 150 generates a character string image by logically synthesizing the character string images in each binarized image, or for each character code of each character corresponding to the character string image in each binarized image The character string image may be generated based on the result of the counting.

FIG. 2 is a flowchart showing an example of processing according to this embodiment.
In step S202, the image reception module 110 receives an image.
In step S204, each binarization module 120 binarizes the image to generate a plurality of binarized images.
In step S206, each starting character search module 130 selects a pixel block (starting character) that is highly likely to be a character from each binarized image.
In step S208, each character string search module 140 searches for a character adjacent to the selected pixel block (starting character) and extracts a character string.
In step S210, the character string synthesis module 150 synthesizes the extracted character string.
In step S212, the character string composition module 150 outputs a character string.

FIG. 3 is a conceptual module configuration diagram of a configuration example in the starting character search module 130 of the present embodiment. The starting character search module 130 includes a character recognition module 310, a complexity calculation module 320, and a starting character search processing module 330.
The character recognition module 310 is connected to the starting character search processing module 330. The character recognition module 310 performs character recognition processing. As the character recognition result, the character code and the certainty factor of the character are output.
The complexity calculation module 320 is connected to the starting character search processing module 330. The complexity calculation module 320 measures the number of alternating turns described above. In addition, the number of strokes of characters recognized by the character recognition module 310 is output. For example, a table that stores character codes and the number of strokes in correspondence with each other may be prepared in advance, the character codes may be searched from the table, and the corresponding number of strokes may be output.
The starting character search processing module 330 is connected to the character recognition module 310 and the complexity calculation module 320. The starting character search processing module 330 selects the pixel block of the starting character based on the information output from the character recognition module 310 and the complexity calculation module 320, which are the possibility that the pixel block in the binarized image is a character. .

FIG. 4 is a flowchart showing an example of processing by the present embodiment (starting character search module 130).
In step S402, the character recognition module 310 performs character recognition processing on the binarized image.
In step S404, the complexity calculation module 320 calculates the complexity of the character.
In step S406, the starting character search processing module 330 selects a starting character in the character string.

FIG. 5 is a conceptual module configuration diagram of a configuration example in the character string search module 140 of the present embodiment.
The character string search module 140 includes a window setting module 510, a character recognition module 520, a selection module 530, and a control module 540.
The window setting module 510 is connected to the control module 540. The window setting module 510 sets a window corresponding to the size of the circumscribed rectangle of the starting character selected by the starting character search module 130. The size may be the same as the size of the circumscribed rectangle of the starting character, or may be a size obtained by multiplying the length and width of the circumscribed rectangle of the starting character by a predetermined magnification. This will be described with reference to the example of FIG. Assume that the character image of “quotient” is selected as the starting character 1110 from the binarized image 900 shown in the example of FIG. 9 by the starting character search module 130. This is because this character is a highly complex character in the binarized image 900. The window setting module 510 sets a window having a size obtained by multiplying the size of the circumscribed rectangle of the starting character 1110 by a predetermined magnification. For example, the search window 1130 is larger than the starting character 1110. Specifically, when the circumscribed rectangle of the starting character 1110 is N × N, the search window 1130 is set to tN × tN. The search window 1130 is scanned so as to make one round around the circumscribed rectangle of the starting character 1110. As scanning, for example, the upper left of the circumscribed rectangle of the starting character 1110 (upper left corner of the search range 1120), upper, upper right (upper right corner of the search range 1120), right, lower right (lower right corner of the search range 1120), lower, lower left The eight search windows 1130 on the left (lower left corner of the search range 1120) may be set, or a predetermined pixel (for example, the upper left corner of the search range 1120) from the upper left corner of the circumscribed rectangle of the starting character 1110 (for example, The search window 1130 may be set so as to move around the circumscribed rectangle of the starting character 1110 by one pixel). Therefore, the search range 1120 is a rectangle of “(2t + 1) N × (2t + 1) N” with the starting character 1110 as the center. T is preferably 1 or more. Furthermore, 1 or more and less than 2 are desirable.

The character recognition module 520 is connected to the control module 540. The character recognition module 520 is equivalent to the character recognition module 310 described above, and one character recognition module may be used by the starting character search module 130 and the character string search module 140. The character recognition module 520 recognizes characters in an image in the window (for example, the search window 1130) set by the window setting module 510.
The control module 540 is connected to the window setting module 510, the character recognition module 520, and the selection module 530. The control module 540 causes the search window 1130 to make a round around the origin character 1110 (for example, search direction 1142, search direction 1144, search direction 1146, search direction 1148), and the pixel block that has entered the search window 1130 in the process. The combination of is calculated.

The moving amount of the search window 1130 may be moved for each pixel block, or may be moved for each predetermined pixel (for example, for each size of the search window 1130, for each pixel, etc.). Also good.
Further, as shown in the example of FIG. 12, the search range 1120 may be set to other binarized images (for example, all binarized images). In other words, the search range 1120 is not set only for the binarized image 900B in which the starting character 1110B is selected, but the search range 1120 is also set in the binarized image 900A and the binarized image 900Z. In the image 900A, the periphery of the starting image (image at a position corresponding to the starting character 1110B) is scanned with the search window 1130A, and in the binarized image 900B, the periphery of the starting character 1110B is scanned with the search window 1130B. Then, the search window 1130Z scans around the origin image (image at a position corresponding to the origin character 1110B).

When the search window 1130 is moved and “TA” is included as shown in the example of FIG. 13, the search window 1330 has four pixel clusters (connected regions) including noise (non-character) 930. . Specifically, in the search window 1330 shown in the example of FIG. 14A, as shown in the example of FIG. 14B, there are four recognition objects 1410, recognition objects 1420, recognition objects 1430, and recognition objects 1440. . Therefore, a total of 2 ⁴ -1 combinations are possible. It should be noted that “−1” in this expression is excluded from the object when there is nothing (when the search window 1330 is blank). The character recognition module 520 causes the character recognition module 520 to recognize characters of each one of these combinations.

  The selection module 530 is connected to the control module 540. The selection module 530 uses, as the second character next to the starting character, a combination of pixel blocks in the search window 1130 whose certainty of the result of character recognition by the character recognition module 520 is higher or higher than the third threshold. select. However, in a plurality of binarized images, when there are a plurality of character candidates exceeding the threshold at the same search window position, the character with the highest certainty is selected.

  The search for adjacent characters performed by the control module 540 will be described with reference to the example of FIG. FIG. 15 is an explanatory diagram showing a processing example according to the present exemplary embodiment (character string search module 140). This search uses recursive iterations. In this example, the first threshold value used for selecting the starting character is different from the third threshold value used for selecting the adjacent character. It is desirable to make the first threshold value higher than the third threshold value. In the example of FIG. 15, the first threshold value is 0.999, and the third threshold value is 0.990. The certainty factor is a value from 0 to 1, and the closer the value is to 1, the higher the certainty factor is (that is, the possibility of the character is high). The character recognition result (quotient) 1510 was selected as the starting character because the certainty factor was 0.9997. Of course, other conditions (complexity, not in a predetermined list of characters) are also met. As a result of scanning around the character recognition result (quotient) 1510, the certainty factor of the character recognition result (ten) 1520 was 0.9953, and the certainty factor of the character recognition result (store) 1550 was 0.9921. . That is, since it is higher than the third threshold, the character recognition result (ta) 1520 and the character recognition result (store) 1550 are selected as the next character after the starting character. Here, the search is performed with depth priority (high priority is given priority). Therefore, the result of scanning around the character recognition result (ta) 1520 using the character recognition result (ta) 1520 as a starting character (note that the character recognition result (quotient) 1510 is excluded from the target), and the character recognition result (ka) 1530. Therefore, the character recognition result (ka) 1530 is selected as the next character after the character recognition result (ta) 1520. Next, the result of scanning around the character recognition result (ka) 1530 using the character recognition result (ka) 1530 as a starting character (note that the character recognition result (ta) 1520 is excluded from the target), the character recognition result (na) Since the certainty factor of 1540 is 0.9974, which is higher than the third threshold, the character recognition result (NA) 1540 is selected as the next character after the character recognition result (KA) 1530. As a result of scanning around the character recognition result (NA) 1540 (note that the character recognition result (KA) 1530 is excluded from the target), there is nothing higher than the third threshold value, so the order returns (character recognition result (NA 1540, character recognition result (ka) 1530, character recognition result (ta) 1520, character recognition result (quotient) 1510 in this order). Next, the result of scanning the vicinity of the character recognition result (store) 1550 (note that the character recognition result (quotient) 1510 is excluded from the target), and since there is nothing higher than the third threshold value, the process ends. As a result, the search order is “quotient → ta → ka → na → store”. Of course, it becomes “Nakata store” when arranged based on the position of the character.

FIG. 6 is a flowchart illustrating a processing example according to the present exemplary embodiment (character string search module 140).
In step S602, the window setting module 510 sets a search window according to the starting character.
In step S604, the character recognition module 520 performs character recognition processing on the image in each search window.
In step S606, the selection module 530 selects a character next to the starting character from the character recognition result.
In step S608, the control module 540 determines whether or not the process has been completed. If the process has been completed, the process ends (step S699). Otherwise, the process proceeds to step S610.
In step S610, the selected character is set as the starting character, and the process returns to step S602.

FIG. 7 is a conceptual module configuration diagram of a configuration example in the character string search module 140 of the present embodiment. The character string search module 140 includes a window setting module 510, an exclusion module 710, a character recognition module 520, a selection module 530, and a control module 540. An exclusion module 710 is added to the character string search module 140 shown in the example of FIG.
The exclusion module 710 is connected to the control module 540. The exclusion module 710 excludes a pixel block combination that matches a predetermined condition from the processing target of the character recognition module 520. As the predetermined condition, for example, one of a combination including pixel blocks of different colors, a combination including a different stroke width, or a combination including a stroke width different from the stroke width of the starting character, Or there are combinations of these. As for the color of the pixel block, the color of the pixel corresponding to the position of the pixel block may be extracted from the image (color image) received by the image receiving module 110. The condition of “combination including pixel clusters of different colors” is to exclude those having different colors within one character. “Different stroke width” means that each pixel block has a different width in the combination of pixel blocks. “Stroke width different from the stroke width of the starting character” means that the width of the circumscribed rectangle of the starting character is different from the width of the circumscribed rectangle surrounding the combination of pixel blocks. Here, “difference” is not limited to a case other than a perfect match, but may be a case where the difference between the minimum and maximum pixel block widths is not within a predetermined range. Therefore, if it is within that range, it is determined that there is no difference.

FIG. 8 is a flowchart showing an example of processing according to this embodiment.
In step S802, the window setting module 510 sets a search window according to the starting character.
In step S804, the exclusion module 710 excludes a foreign search window image from the character recognition target.
In step S806, the character recognition module 520 performs character recognition processing on the image in the search window.
In step S808, the selection module 530 selects a character next to the starting character from the character recognition result.
In step S810, the control module 540 determines whether or not the process has been completed. If the process has been completed, the process ends (step S899). Otherwise, the process proceeds to step S812.
In step S812, the selected character is used as a starting character, and the process returns to step S802.
In step S806, the character recognition process is not performed on all the images in the search window, but the character recognition process is performed on the images in the search windows other than the search window excluded in step S804.

  FIG. 16 is an explanatory diagram showing a processing example according to the present embodiment (character string synthesis module 150). A character string search result 1610 and a character string search result 1620, which are character string images in each binarized image, are combined to generate a character string combination result 1690. In the example of FIG. 16, the result of logical sum (OR) synthesis is shown. Note that only the character string search result 1610, the character string search result 1620, and the result of the two character strings are shown, but only the number of starting characters is generated. In addition, the character code of each character corresponding to the character string image in each binarized image may be counted, and the character string image may be generated based on the result of the count. That is, the character string image is generated by selecting the image corresponding to the character code having the largest number as a result of the majority of the character codes of each character.

  A hardware configuration example of the image processing apparatus according to the present embodiment will be described with reference to FIG. The configuration shown in FIG. 17 is configured by a personal computer (PC), for example, and shows a hardware configuration example including a data reading unit 1717 such as a scanner and a data output unit 1718 such as a printer.

  A CPU (Central Processing Unit) 1701 includes various modules described in the above-described embodiments, that is, the image reception module 110, the binarization module 120, the starting character search module 130, the character string search module 140, and a character string synthesis module. 150, a character recognition module 310, a complexity calculation module 320, a starting character search processing module 330, a window setting module 510, a character recognition module 520, a selection module 530, a control module 540, an exclusion module 710, and the like. The control unit executes processing according to the computer program.

  A ROM (Read Only Memory) 1702 stores programs used by the CPU 1701, calculation parameters, and the like. A RAM (Random Access Memory) 1703 stores programs used in the execution of the CPU 1701, parameters that change as appropriate in the execution, and the like. These are connected to each other by a host bus 1704 including a CPU bus.

  The host bus 1704 is connected to an external bus 1706 such as a PCI (Peripheral Component Interconnect / Interface) bus via a bridge 1705.

  A keyboard 1708 and a pointing device 1709 such as a mouse are input devices operated by an operator. The display 1710 includes a liquid crystal display device or a CRT (Cathode Ray Tube), and displays various types of information as text or image information.

  An HDD (Hard Disk Drive) 1711 has a built-in hard disk, drives the hard disk, and records or reproduces a program executed by the CPU 1701 and information. The hard disk stores an image received by the image receiving module 110, a binarized image layer 122, information about a starting character, a character recognition result, a character string (character string image) that is a final processing result, and the like. Further, various computer programs such as various other data processing programs are stored.

  The drive 1712 reads out data or a program recorded on a removable recording medium 1713 such as a magnetic disk, an optical disk, a magneto-optical disk, or a semiconductor memory, and the data or program is read out as an interface 1707 and an external bus 1706. , The bridge 1705, and the RAM 1703 connected via the host bus 1704. The removable recording medium 1713 can also be used as a data recording area similar to the hard disk.

  The connection port 1714 is a port for connecting the external connection device 1715 and has a connection unit such as USB or IEEE1394. The connection port 1714 is connected to the CPU 1701 and the like via an interface 1707, an external bus 1706, a bridge 1705, a host bus 1704, and the like. The communication unit 1716 is connected to a communication line and executes data communication processing with the outside. The data reading unit 1717 is a scanner, for example, and executes document reading processing. The data output unit 1718 is a printer, for example, and executes document data output processing.

  Note that the hardware configuration of the image processing apparatus illustrated in FIG. 17 illustrates one configuration example, and the present embodiment is not limited to the configuration illustrated in FIG. 17, and the modules described in the present embodiment are executed. Any configuration is possible. For example, some modules may be configured with dedicated hardware (for example, Application Specific Integrated Circuit (ASIC), etc.), and some modules are in an external system and connected via a communication line In addition, a plurality of systems shown in FIG. 17 may be connected to each other via communication lines so as to cooperate with each other. Further, it may be incorporated in a copying machine, a fax machine, a scanner, a printer, a multifunction machine (an image processing apparatus having any two or more functions of a scanner, a printer, a copying machine, a fax machine, etc.).

  Further, in the description of the above-described embodiment, “more than”, “less than”, “greater than”, and “less than (less than)” in a comparison with a predetermined value contradicts the combination. As long as the above does not occur, “larger”, “smaller (less than)”, “more”, and “less” may be used.

The program described above may be provided by being stored in a recording medium, or the program may be provided by communication means. In that case, for example, the above-described program may be regarded as an invention of a “computer-readable recording medium recording the program”.
The “computer-readable recording medium on which a program is recorded” refers to a computer-readable recording medium on which a program is recorded, which is used for program installation, execution, program distribution, and the like.
The recording medium is, for example, a digital versatile disc (DVD), which is a standard established by the DVD Forum, such as “DVD-R, DVD-RW, DVD-RAM,” and DVD + RW. Standard “DVD + R, DVD + RW, etc.”, compact disc (CD), read-only memory (CD-ROM), CD recordable (CD-R), CD rewritable (CD-RW), Blu-ray disc ( Blu-ray (registered trademark) Disc), magneto-optical disk (MO), flexible disk (FD), magnetic tape, hard disk, read-only memory (ROM), electrically erasable and rewritable read-only memory (EEPROM (registered trademark)) )), Flash memory, Random access memory (RAM) SD (Secure Digital) memory card and the like.
The program or a part of the program may be recorded on the recording medium for storage or distribution. Also, by communication, for example, a local area network (LAN), a metropolitan area network (MAN), a wide area network (WAN), a wired network used for the Internet, an intranet, an extranet, etc., or wireless communication It may be transmitted using a transmission medium such as a network or a combination of these, or may be carried on a carrier wave.
Furthermore, the program may be a part of another program, or may be recorded on a recording medium together with a separate program. Moreover, it may be divided and recorded on a plurality of recording media. Further, it may be recorded in any manner as long as it can be restored, such as compression or encryption.

DESCRIPTION OF SYMBOLS 100 ... Image processing apparatus 110 ... Image reception module 120 ... Binarization module 122 ... Binary image layer 130 ... Origin character search module 140 ... Character string search module 150 ... Character string synthesis module

Claims (10)

A plurality of binarization means for binarizing the image;
Selection means for selecting a pixel block based on a possibility that the pixel block in each binarized image binarized by the binarization unit is a character;
Extraction means for searching for adjacent characters from the pixel block selected by the selection means and extracting a character string;
An image processing apparatus comprising: a combining unit that combines the character strings extracted by the extracting unit. The selection means has a certainty factor of a character as a result of character recognition of each of the binarized images higher than or equal to a predetermined first threshold, and the complexity of the character is higher than a second threshold. The image processing apparatus according to claim 1, wherein the pixel block is selected from the above. The image processing apparatus according to claim 2, wherein the complexity is any one of a number of alternations in a circumscribed rectangle of the pixel block, a number of strokes of characters as a character recognition result, or a combination thereof. . The selection means is a pixel block in which the certainty factor of the character as a result of character recognition of each of the binarized images is higher or higher than a predetermined first threshold value, and the character is not a predetermined character. The image processing apparatus according to claim 1, wherein the image processing apparatus is selected. The extraction unit sets a rectangle corresponding to the size of a circumscribed rectangle of the first pixel block selected by the selection unit, and performs character recognition in the rectangle of each binarized image around the circumscribed rectangle. Then, a combination of pixel blocks whose certainty of the result of character recognition is higher than or higher than a third threshold is set as a second character next to the first character, and then the second character is The image processing apparatus according to any one of claims 1 to 4, wherein the character string is extracted by setting the first character. The image processing apparatus according to claim 5, wherein the extraction unit performs the character recognition by excluding a combination of pixel blocks matching a predetermined condition. The predetermined condition is a combination including a pixel block of a different color, a combination including a different stroke width, or a combination including a stroke width different from the stroke width of the character in the first pixel block. The image processing apparatus according to claim 6, wherein the image processing apparatus is any one of the above or a combination thereof. The synthesizing unit generates a character string image by performing a logical sum synthesis on the character string images in the binarized images, or for each character code of each character corresponding to the character string image in the binarized images. The image processing apparatus according to claim 1, wherein a character string image is generated based on a result of the counting. The binarization unit performs binarization using a plurality of different thresholds, binarization using a plurality of different thresholds after inversion of the image, or binarization of a combination thereof. The image processing apparatus according to any one of the above. Computer
A plurality of binarization means for binarizing the image;
Selection means for selecting a pixel block based on a possibility that the pixel block in each binarized image binarized by the binarization unit is a character;
Extraction means for searching for adjacent characters from the pixel block selected by the selection means and extracting a character string;
An image processing program for causing a character string extracted by the extracting unit to function as a combining unit.

Images