JP5489894B2 - Image processing apparatus and image processing program - Google Patents

Description

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

  In Patent Document 1, simple features that differ depending on the type or handwritten character are extracted from character data of one character, and it is determined with high accuracy whether the target character is a typed character or a handwritten character. As a task, the feature extraction unit determines whether the character is a printed character or a handwritten character from the character data of one character extracted by the character data extraction unit (direction of character stroke, thick character). The character type determination unit determines whether the character type is a printed character or a handwritten character based on the feature, and according to the determination result, the character type recognition unit, the handwritten character recognition unit It is disclosed that either of the above is executed.

  In Patent Document 2, it is an object to make it possible to identify handwritten characters and printed characters with high accuracy in a short time, and to shorten the time required for character recognition processing. Characters from image data read from a document or a form are disclosed. A pattern is extracted, and a handwritten / print type discriminating unit discriminates a handwritten character and a type. The handwriting / type discriminating unit includes at least a feature representing the complexity of the character pattern and a feature representing linearity from the character pattern. N feature vectors are obtained, and using the feature vectors, it is determined whether the character pattern is a handwritten character or a printed character. In the character recognition unit, depending on whether the character data is a handwritten character or a printed character, It is disclosed that corresponding character recognition processing can be performed, and line width variation, character position variation, and the like can be used as characteristics of the character pattern.

  In Patent Document 3, an optical that can quickly determine whether a character to be read is a handwritten character or a printed character, and can improve the reading speed of a form described in a mixture of handwritten characters and printed characters. The field cutout unit cuts out character data in field units from the character data read by the scanner in the sheet buffer and stores it in the field buffer. The character type determination unit The character type of the character data in the field buffer is determined, and the determination result is notified to the recognition unit. The recognition unit uses either the handwritten dictionary or the type dictionary according to the determination result to determine the character data in the field buffer. Performing character recognition is disclosed.

  In Patent Document 4, it is an object to improve the character discrimination accuracy by using features other than the height of character data and using a plurality of features together. The rectangle (black pixel concatenation rectangle) is extracted, the character rectangle feature calculation unit calculates the area of each extracted rectangle, the standard deviation calculation unit calculates the standard deviation of the set of rectangular area values, and the comparison determination unit calculates the standard deviation. It is disclosed that when a predetermined threshold value is exceeded, it is determined that the character in the predetermined area is a handwritten character, and when it is smaller than the threshold value, it is determined that it is a printed character, and recognition processing is performed using each dictionary. Yes.

  In Patent Document 5, it is an object to be able to determine the character type efficiently and with high accuracy, and to execute the determination process at a high speed, and a plurality of simple conditions, which are printed character strings, are combined in descending order of probability to meet each condition. Whether it is a printed character example or a handwritten character string is disclosed.

Japanese Patent Laid-Open No. 10-162102 JP 2000-181993 A JP 05-189604 A JP-A-10-214308 JP 2000-339404 A

  The present invention relates to an image processing apparatus that determines whether a character in a character string image is a type or a handwriting by extracting information on the angle, length, and appearance frequency of the line segment from the character string image. And an image processing program.

The gist of the present invention for achieving the object lies in the inventions of the following items.
According to the first aspect of the present invention, fine line conversion means for converting a character string image including a character string into thin line information, and area setting means for setting a rectangular area in the character string image that has become a thin line by the thin line conversion means. And Hough transforming means for performing a Hough transform on the rectangular region set by the region setting means, angle information of a line segment as a result of the Hough transform processing by the Hough transform means, and height information thereof A feature extracting unit for extracting, and a determining unit for determining whether a character string in the character string image is a printed character or a handwritten character based on the feature extracted by the feature extracting unit. An image processing apparatus.

  According to a second aspect of the present invention, the feature extraction unit weights the processing result of each rectangle of the Hough transform by the Hough transform unit, and removes the weight that is determined not to be line segment information A first means for extracting a first value from the processing result by comparing the processing result by the adding means with a distribution in angle with a predetermined value; And a first feature quantity extraction means for extracting, as a feature quantity, a difference between the value extraction means and the first value extracted by the first value extraction means and a predetermined value. An image processing apparatus according to claim 1.

  According to a third aspect of the present invention, the feature extraction unit weights the processing result of each rectangle of the Hough transform by the Hough transform unit, and removes the weight determined to be not line segment information A first means for extracting a first value from the processing result by comparing the processing result by the adding means with a distribution in angle with a predetermined value; A second feature amount extraction unit that extracts, as a feature amount, a variance of a processing result by the addition unit within a predetermined range with respect to the first value extracted by the first value extraction unit; The image processing apparatus according to claim 1, further comprising: means.

  According to a fourth aspect of the present invention, the feature extraction unit uses, as a feature amount, a variance of a processing result by the addition unit within a predetermined range with respect to the first value extracted by the first value extraction unit. The image processing apparatus according to claim 2, further comprising second feature amount extraction means for extracting.

  According to a fifth aspect of the present invention, the feature extraction unit extracts a second value from the processing result by comparing the processing result of each rectangle of the Hough transform by the Hough transform unit with a predetermined value. And second feature extraction means for extracting, as feature quantities, the height and angle of the second value extracted by the second value extraction means. Item 5. The image processing device according to any one of Items 1 to 4.

  According to a sixth aspect of the present invention, the feature extraction unit calculates a variance of the processing result of each rectangle of the Hough transform performed by the Hough transform unit within a predetermined range with respect to a predetermined value. And a fourth feature amount extracting means for extracting the total and height of the variances of the respective rectangles calculated by the variance calculating means as feature amounts. The image processing apparatus according to the item.

  According to the seventh aspect of the present invention, the computer sets a rectangular area in the character string image that has become a thin line by the thin line conversion means, and fine line conversion means for converting the character string image including the character string into thin line information. Area setting means, Hough transform means for performing Hough transform on the rectangular area set by the area setting means, angle information of the line segment of the result of Hough transform processing by the Hough transform means, and height information thereof And a feature extraction unit that extracts a character string as a feature, and a determination unit that determines whether a character string in the character string image is a type or a handwriting based on the feature extracted by the feature extraction unit. This is a featured image processing program.

  According to the image processing device of claim 1, by extracting information on the angle and length of the line segment of the character and the appearance frequency from the character string image, the character in the character string image is either printed or handwritten. Can be determined.

  According to the image processing apparatus of the second aspect, it is possible to determine whether the type is a type or a handwriting by using a difference between the first value and a predetermined value as a feature amount.

  According to the image processing apparatus of the third aspect, it is possible to determine whether the character is handwritten or handwritten using a variance within a predetermined range with respect to the first value as a feature amount.

  According to the image processing apparatus of claim 4, the difference between the first value and a predetermined value, or the variance within a predetermined range with respect to the first value is used as a feature value in either type or handwriting. It can be determined whether there is.

  According to the image processing device of the fifth aspect, it is possible to determine whether the character is a type or a handwriting by using the height and angle of the second value as the feature amount.

  According to the image processing apparatus of the sixth aspect, it is possible to determine whether the character is handwritten or handwritten by using the total and height of the variance of the processing result of the Hough transform as a feature amount.

  According to the image processing program of claim 7, by extracting information on the angle, length, and appearance frequency of the character line segment from the character string image, the character in the character string image is either printed or handwritten. Can be determined.

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 explanatory drawing which shows the example of the character string image extracted by the character string extraction module. It is explanatory drawing which shows the example of the result thinned by the thinning process module. It is explanatory drawing which shows the example of character height. It is explanatory drawing which shows the example of the rectangle which divides | segments a character string image by the rectangle division | segmentation process module. It is explanatory drawing which shows the example of the process result of the Hough conversion by a Hough conversion process module. It is explanatory drawing which shows the example of the result of the threshold value process by a threshold value processing module. It is a conceptual module block diagram (1) about the structural example in a feature extraction module. It is a flowchart which shows the process example by a feature extraction module. It is explanatory drawing which shows the example of a data structure of a teacher data table. It is explanatory drawing which shows the example of the result of the addition process by an addition process module. It is explanatory drawing which shows the example of a process of a peak and teacher data comparison module. It is a conceptual module block diagram (2) about the structural example in a feature extraction module. It is a flowchart which shows the process example by a feature extraction module. It is explanatory drawing which shows the process example by a peak extraction module. It is explanatory drawing which shows the process example by a peak feature extraction module. It is a conceptual module block diagram (3) about the structural example in a feature extraction module. It is a flowchart which shows the process example by a feature extraction module. It is explanatory drawing which shows the example of a process by a dispersion | distribution total and height extraction module. 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.
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 according to the present embodiment determines whether a character string in an image is printed or handwritten. As shown in the example of FIG. 1, the image receiving module 110, character string extraction Module 120, thinning processing module 130, rectangle size determination module 140, rectangle division processing module 150, Hough transform processing module 160, threshold processing module 170, feature extraction module 180, distance measurement module 185, determination module 190, and output module 195 Have.

  The image reception module 110 is connected to the character string extraction module 120, receives an image, and passes the image to the character string extraction module 120. Accepting an image means, for example, reading an image with a scanner, a camera, etc., receiving an image from an external device via a communication line by fax, taking a video with a CCD (Charge-Coupled Device), etc. This includes 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 may be a binary image or a multi-value image (including a color image). One image may be received or a plurality of images may be received. In addition, as long as the content of the image includes a character string, it may be a document used for business, a pamphlet for advertisement, or the like.

The character string extraction module 120 is connected to the image reception module 110 and the thinning processing module 130, receives an image from the image reception module 110, extracts a character string image including a character string, and converts the character string image into a thin line. To the processing module 130. Extraction of the character string image uses existing techniques such as structural analysis. For example, a blank having a predetermined width or more is detected from an image, divided into regions by the blank, and the type of the divided region (character region, graphic region, photo region, etc.) is distributed in the size of the pixel block The coordinates of the character area (for example, the upper left X coordinate, Y coordinate, and the width and height of the area) are extracted. Then, the coordinates of the character string image (for example, the upper left X coordinate, Y coordinate, and the width and height of the character string image) are extracted from the character area. For example, the pixels may be counted in the horizontal direction and the vertical direction in the character area, and the regularly arranged column may be determined as the character string.
Note that the pixel block includes at least a pixel area that is continuous in four or eight connections, and includes a set of these pixel areas. 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 is, for example, an image area in which the pixel areas are close to each other in distance, an image area that is projected vertically or horizontally so as to cut out one character at a time from a line as a sentence, and cut out at a blank spot Or an image region cut out at regular intervals.
Further, the character string here may be a character string in which characters are arranged in the horizontal direction, or may be a character string in which characters are arranged in the vertical direction.

  The thinning processing module 130 is connected to the character string extraction module 120 and the rectangular size determination module 140, receives a character string image from the character string extraction module 120, converts the character string image into thin line information, The thinned character string image is passed to the rectangular size determination module 140. The thinning process for converting into thin line information uses existing technology. For example, thinning may be performed using a filter that defines pixels to be removed. Moreover, you may make it extract a contour line. The thinning process here refers to converting an image into information of only a line segment that is not affected by the thickness of characters.

The rectangle size determination module 140 is connected to the thinning processing module 130 and the rectangle division processing module 150, determines the size of a rectangle that divides the character string image thinned by the thinning processing module 130, and the rectangle. The character string image that has been subjected to the thinning process along with the size of the image is passed to the rectangular division processing module 150. The rectangle may be a rectangle or a square. The size of the rectangle to be determined is the height and width in the case of a rectangle, and the length of one side in the case of a square.
The size of the rectangle may be, for example, the size of characters in the character string image. In this case, the size of the rectangle may be determined from the size of the pixel block in the character string image. For example, when a pixel block is extracted as described above, an image of one character increases as one pixel block. In the case of a character string image that is long in the horizontal direction, it may be a square with the height of the character in the character string image as one side. In the case of a character string image that is long in the vertical direction, it may be a square with the width of the character in the character string image as one side. Further, a value obtained by adding a predetermined value (a value of 0 or more) to the size of the pixel block, the height of the character, or the like may be used as the rectangular size.

The rectangular division processing module 150 is connected to the rectangular size determination module 140 and the Hough conversion processing module 160, and the character string image thinned by the thinning processing module 130 is determined by the rectangular size determination module 140. The image is divided into rectangles, and the divided rectangle image is passed to the Hough transform processing module 160. Here, the division of the rectangle may be performed in order from the end for each size of the rectangle, or may be performed at a position where pixels do not exist as much as possible (that is, a character (with one character within the rectangle). May be divided so that they can be included), or may be divided so that the rectangles overlap (overlapping) (that is, the overlapped portion contains the same image) (This corresponds to the case where the line segment breaks at the boundary of the rectangle), and there may be a distance between rectangles (that is, the rectangle is set apart from the adjacent rectangle) To do).
Therefore, the rectangular size determination module 140 and the rectangular division processing module 150 set a rectangular area in the character string image that has been thinned by the thinning processing module 130.

  The Hough conversion processing module 160 is connected to the rectangular division processing module 150 and the threshold processing module 170, performs Hough conversion on the rectangular image divided by the rectangular division processing module 150, and the processing result To the threshold processing module 170. The Hough transform is to detect a straight line, and calculates a pair of a distance (length of a normal line drawn on the straight line) and an angle from the center (origin) to the target straight line and counts it. The Hough transform processing result is a set of distance and angle from the center (origin) to the target straight line, and a counting result (hereinafter also referred to as height) of the set. For example, as shown in FIG. 7, it can be expressed on a Hough space constituted by an axis of distance from the center to a straight line, an axis of angle, and an axis of height.

The threshold processing module 170 is connected to the Hough conversion processing module 160 and the feature extraction module 180, and compares the height of the Hough conversion processing result by the Hough conversion processing module 160 with a predetermined threshold. The processing result is selected, and the selected processing result is passed to the feature extraction module 180. The selection process here is a process for excluding those determined not to be line segment information. For example, a feature amount whose height is greater than the threshold (corresponding to the above-described selected processing result, which is also referred to as a rectangular feature amount) is extracted as direction information of a line segment constituting the character. Further, this threshold value may be set based on the size of the rectangle determined by the rectangle size determination module 140. For example, a table storing threshold values in association with the size of a rectangle is prepared in advance, and the threshold value is set from the size of the rectangle using this table. Further, the threshold may be set from the size of the rectangle based on a function that increases the threshold as the size of the rectangle increases. In addition to selecting the processing result as it is, selecting the processing result may be performed by subtracting the threshold value from the height of the processing result and deleting the result that is 0 or less. As a result, the processing result whose height is reduced by the threshold value is selected. Note that the threshold processing module 170 may not be provided. That is, the Hough conversion processing result by the Hough conversion processing module 160 may be passed to the feature extraction module 180.
Further, the threshold processing module 170 may weight each rectangular processing result of the Hough conversion performed by the Hough conversion processing module 160. And you may make it select the process result which weighted by the selection process mentioned above. In addition, this process is performed when the process by the addition process module 910 mentioned later is performed.

  The feature extraction module 180 is connected to the threshold processing module 170 and the distance measurement module 185, extracts a feature based on the angle of the processing result selected by the threshold processing module 170, and sends the extracted feature to the distance measurement module 185. hand over. A module configuration example, a processing example, and the like in the feature extraction module 180 will be described later with reference to FIGS. When there is no threshold processing module 170, the processing result of the Hough conversion by the Hough conversion processing module 160 is targeted. Then, the angle information and the height information of the line segment of the processing result are extracted as features. Note that the height information as the processing result of the Hough transform represents the length of the line segment and the likelihood of the line segment.

The distance measurement module 185 is connected to the feature extraction module 180 and the determination module 190, and measures the distance between the type feature in the feature space and the feature extracted by the feature extraction module 180. The feature of the type here is so-called teacher data, which is created by extracting the feature of the type in advance.
The determination module 190 is connected to the distance measurement module 185 and the output module 195. Based on the distance measured by the distance measurement module 185, it is determined whether the character string in the character string image is printed or handwritten. The determination result is passed to the output module 195. For example, if the distance measured by the distance measurement module 185 is within a predetermined range, it is determined to be a printed character, and otherwise it is determined to be a handwritten character.

The output module 195 is connected to the determination module 190 and outputs a determination result by the determination module 190. The output destination may be a storage device or another information processing apparatus via a communication line. For example, the following processing is performed as an output destination information processing apparatus.
(1) When character recognition is performed on a document in which printed characters and handwritten characters are mixed, the processing according to the present embodiment is performed, and the printed character portion is a printed character recognition device, and the handwritten character portion is a handwritten character recognition device. Is used.
(2) The process according to the present embodiment is performed to determine the part added by handwriting, erase the part, and reproduce the original image.
(3) A document added by handwriting is retrieved from the document image.

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, the character string extraction module 120 performs structural analysis and extracts a character string image. For example, the character string image 300 shown in FIG. 3 is extracted. In this example, an image of a horizontally written character string is extracted.
In step S206, the thinning processing module 130 performs thinning processing on the character string image. For example, the thinned image 400 shown in FIG. 4 is the result of thinning the character string image 300 shown in FIG. A thick line segment is converted into a 1-dot line segment.

In step S208, the rectangle size determination module 140 determines the rectangle size. For example, the rectangle size is determined with the character height 510 of the character image in the thinned image 400 shown in FIG. 5 as the size of one side of the rectangle (square).
In step S210, the rectangular division processing module 150 divides the character string image into rectangular images using the rectangular size. FIG. 6 is an explanatory diagram illustrating an example of a rectangle in which a character string image is divided by the rectangle division processing module 150. In this example, the thinned image 400 is divided into eight rectangular images of rectangles 611 to 618. Here, the rectangles are divided in order so as not to overlap. Further, the rectangle is not limited to one character, and may include a plurality of characters (typically two characters).

In step S212, the Hough conversion processing module 160 performs a Hough conversion process on each rectangular image. FIG. 7 is an explanatory diagram illustrating an example of the result of the Hough conversion processing performed by the Hough conversion processing module 160. This is a graph representation of the processing result of the Hough transform for the rectangle 611 illustrated in FIG.
In step S214, the threshold processing module 170 performs threshold processing on the processing result of the Hough transform. FIG. 8 is an explanatory diagram illustrating an example of a result of threshold processing by the threshold processing module 170. This is obtained by performing threshold processing on the processing result of the Hough transform exemplified in FIG. That is, a processing result having a height higher than a predetermined threshold is extracted. As described above, step S214 is not necessary.

In step S216, the feature extraction module 180 performs feature extraction processing.
In step S218, the distance measurement module 185 measures the distance between the feature extracted in step S216 and the teacher data of the type.
In step S220, the determination module 190 determines whether the image of the target character string is a handwritten character or a printed character.
In step S222, the output module 195 outputs a determination result.

FIG. 9 is a conceptual module configuration diagram (1) for a configuration example in the feature extraction module 180.
The feature extraction module 180 includes an addition processing module 910, a histogram creation processing module 920, a peak extraction module 930, a peak processing module 940, and a feature amount storage module 950.

The addition processing module 910 is connected to the histogram creation processing module 920 and adds the processing results of each rectangle selected by the threshold processing module 170. Here, adding the processing results means adding the heights of the processing results having the same distance and angle from the center to the straight line.
The histogram creation processing module 920 is connected to the addition processing module 910 and the peak extraction module 930, and creates a distribution (hereinafter also referred to as a histogram) in the angle of the processing result by the addition processing module 910. For example, the graph is as shown in FIG. This counts the number of processing results of the Hough transform remaining by the processing by the addition processing module 910 with respect to the angle axis in the Hough space. The counted result is the frequency at that angle.
The peak extraction module 930 is connected to the histogram creation processing module 920 and the peak processing module 940, and compares the frequency in the histogram created by the histogram creation processing module 920 with a predetermined value to obtain the processing result. A first value is extracted from the inside. Hereinafter, the first value is referred to as a peak value. For example, an angle indicating the frequency of a value larger than a predetermined value is extracted as a peak value.

The peak processing module 940 is connected to the peak extraction module 930 and the feature amount storage module 950, and includes a peak / teacher data comparison module 942 and a peak surrounding variance calculation module 944.
The peak / teacher data comparison module 942 extracts the difference between the peak value extracted by the peak extraction module 930 and a predetermined value as a feature amount. As the predetermined value, for example, as in the teacher data table 1100 illustrated in FIG. 11, the processing result (the image processing module 110 to the threshold processing module 170, the addition processing) Results of processing from the module 910 to the peak / teacher data comparison module 942), or values such as 90 degrees, 180 degrees, and 45 degrees without using the processing results for the teacher data. May be. These are angles of lines constituting characters that frequently appear in the character image of the type. Therefore, when this difference is small, it indicates that the possibility of printed characters is high, and when this difference is large, it indicates that the possibility of handwritten characters is high. When the determination module 190 uses this feature amount (difference), it is determined whether the character string in the character string image is printed or handwritten by comparing this difference with a predetermined value.

The peak circumference variance calculation module 944 calculates the variance of the processing result by the addition processing module 910 within a predetermined range (hereinafter also referred to as the circumference of the peak value) with respect to the peak value extracted by the peak extraction module 930. Extract as The predetermined range is, for example, within a predetermined distance from the peak value with the peak value as the center. This feature amount indicates the degree of line direction coincidence. When the variance within this range is small, it indicates that the possibility of printing is high, and when this difference is large, it indicates that the possibility of handwritten characters is high. This is because in the case of printed characters, there is little blur because it is a predetermined character, but blurring occurs if it is a handwritten character. When the determination module 190 uses this feature amount (dispersion), the character string in the character string image is determined to be a printed character or a handwritten character by comparing the distribution with a predetermined value.
Note that either one or both of the peak / teacher data comparison module 942 and the peak circumference dispersion calculation module 944 may be provided.
The feature amount storage module 950 is connected to the peak processing module 940 and stores the feature amount extracted by one or both of the peak / teacher data comparison module 942 and the peak surrounding variance calculation module 944 in the storage device. . The distance measurement module 185 and the determination module 190 use the feature values stored in this storage device.

FIG. 10 is a flowchart illustrating an example of processing performed by the feature extraction module 180.
In step S1002, the addition processing module 910 adds the Hough transform processing results subjected to threshold processing in each rectangle. FIG. 12 is an explanatory diagram showing an example of the result of the addition processing by the addition processing module 910. The heights of the processing results obtained by the Hough conversion processing module 160 and the threshold processing module 170 are added to the images of the rectangles 611 to 618 illustrated in FIG.
In step S1004, the histogram creation processing module 920 creates a histogram for the angle. FIG. 13 is an explanatory diagram showing an example of processing of the peak / teacher data comparison module 942. The frequency distribution in the angle of the addition result illustrated in FIG. 12 is generated.

In step S1006, the peak extraction module 930 extracts the peak value in the histogram. In the example shown in FIG. 13, the vicinity of 0 degrees, 90 degrees, and 180 degrees is extracted as a peak value.
In step S1008, the peak / teacher data comparison module 942 calculates the difference by comparing the peak value with the teacher data.

In step S1010, the peak circumference variance calculation module 944 calculates the variance around the peak value.
In step S1012, the feature amount storage module 950 stores the feature amount.
Note that either one of the processes in steps S1008 and S1010 may be performed. In addition, either may be performed first, or processing may be performed in parallel.

FIG. 14 is a conceptual module configuration diagram (2) of the configuration example in the feature extraction module 180.
The feature extraction module 180 includes a peak extraction module 1410, a peak feature extraction module 1420, and a feature amount storage module 1430.

The peak extraction module 1410 is connected to the peak feature extraction module 1420, and by comparing the processing result (rectangular feature amount) of each rectangle selected by the threshold processing module 170 with a predetermined value, A second value is extracted from the processing result. Hereinafter, the second value is referred to as a peak value. For example, a peak value that is higher than a predetermined value is extracted as a peak value.
The peak feature extraction module 1420 is connected to the peak extraction module 1410 and the feature amount storage module 1430, and extracts the height and angle of the peak value extracted by the peak extraction module 1410 as the feature amount. When the determination module 190 uses this feature amount (a pair of peak value height and angle), a comparison between the distribution of the peak value height and angle pair and a predetermined value is performed to determine whether or not the character string image includes It is determined whether the character string is printed or handwritten. It is used that the dispersion is small if it is a type, and the dispersion is large due to blurring if it is a handwritten character.
The feature quantity storage module 1430 is connected to the peak feature extraction module 1420, and stores the feature quantity extracted by the peak feature extraction module 1420 in a storage device. The distance measurement module 185 and the determination module 190 use the feature values stored in this storage device.

FIG. 15 is a flowchart illustrating an example of processing performed by the feature extraction module 180.
In step S1502, the peak extraction module 1410 extracts a peak value from the Hough transform processing result subjected to threshold processing in each rectangle. FIG. 16 is an explanatory diagram showing an example of processing by the peak extraction module 1410. In the result of the threshold processing illustrated in FIG. 8, three peaks 1612, 1614, and 1616 are extracted.
In step S1504, the peak feature extraction module 1420 extracts the height and angle of the peak value as the rectangular feature. FIG. 17 is an explanatory diagram illustrating an example of processing performed by the peak feature extraction module 1420. The height and angle of the peak value extracted as the feature amount are mapped in a two-dimensional space of height and angle.
In step S1506, the feature amount storage module 1430 stores the feature amount.

FIG. 18 is a conceptual module configuration diagram (3) of the configuration example in the feature extraction module 180.
The feature extraction module 180 includes a variance calculation module 1810, a variance total / height extraction module 1820, and a feature quantity storage module 1830.

  The variance calculation module 1810 is connected to the variance total / height extraction module 1820 and is executed by the threshold processing module 170 within a predetermined range with respect to a predetermined value (hereinafter also referred to as the periphery of the representative angle). The variance of the Hough transform processing result (rectangular feature value) of each selected rectangle is calculated. Regarding the angle, the variance in the angular direction of the data around the representative angle (horizontal (0 degree, 180 degree), vertical (90 degree), italic angle (45 degree, 135 degree, etc.)) of the teacher data of the type is calculated.

The variance total / height extraction module 1820 is connected to the variance calculation module 1810 and the feature amount storage module 1830, and extracts the total variance and height of each rectangle calculated by the variance calculation module 1810 as a feature amount. As described above, since there are a plurality of representative angles of the teacher data of the type, the variances around the representative angle are summed up. Further, as the height, an average height, a mode value, or the like of the rectangular feature amount that is a target of dispersion may be used.
When the determination module 190 uses this feature amount (a combination of the total variance and the height), the combination of the total variance and the height is mapped in a two-dimensional space, and the positional relationship with a predetermined threshold value is used. It is determined whether the character string in the character string image is printed or handwritten. FIG. 20 is an explanatory diagram showing an example of processing by the variance total / height extraction module 1820. The feature values of each rectangle are mapped in a two-dimensional space in which the horizontal axis represents the total variance of the rectangular feature values around the representative angle of the teacher data and the vertical axis represents the height. On the upper side of the threshold 2010 (type region 2020), since there is no blur in the straight lines constituting the character, the peak value tends to be high, and there is no blur in the angle, so the variance for the teacher data is also small. . On the lower side of the threshold 2010 (handwritten character region 2030), the straight lines constituting the characters are blurred compared to the printed characters, the peak value tends to be low, and the characters are perpendicular, horizontal, etc. Since it is hard to become, the dispersion | distribution with respect to teacher data becomes large.

  The feature quantity storage module 1830 is connected to the variance total / height extraction module 1820 and stores the feature quantity extracted by the variance total / height extraction module 1820 in a storage device. The distance measurement module 185 and the determination module 190 use the feature values stored in this storage device.

FIG. 19 is a flowchart illustrating an example of processing performed by the feature extraction module 180.
In step S1902, the variance calculation module 1810 calculates the variance around the teacher data.
In step S1904, the variance total / height extraction module 1820 extracts the variance total and height as features of the rectangle.
In step S1906, the feature amount storage module 1830 stores the feature amount.

  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. 21 is configured by a personal computer (PC), for example, and shows a hardware configuration example including a data reading unit 2117 such as a scanner and a data output unit 2118 such as a printer.

  The CPU 2101 includes various modules described in the above-described embodiments, that is, the character string extraction module 120, the thinning processing module 130, the rectangular size determination module 140, the rectangular division processing module 150, the Hough conversion processing module 160, and the threshold processing module. 170 is a control unit that executes processing according to a computer program describing an execution sequence of each module such as 170, feature extraction module 180, distance measurement module 185, and determination module 190.

  A ROM (Read Only Memory) 2102 stores programs used by the CPU 2101, calculation parameters, and the like. The RAM 2103 stores programs used in the execution of the CPU 2101, parameters that change as appropriate during the execution, and the like. These are connected to each other by a host bus 2104 including a CPU bus.

  The host bus 2104 is connected to an external bus 2106 such as a PCI (Peripheral Component Interconnect / Interface) bus via a bridge 2105.

  A keyboard 2108 and a pointing device 2109 such as a mouse are input devices operated by an operator. The display 2110 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) 2111 includes a hard disk, drives the hard disk, and records or reproduces a program executed by the CPU 2101 and information. The hard disk stores received images, character string images, result data of the feature extraction module 180, and the like. Further, various computer programs such as various other data processing programs are stored.

  The drive 2112 reads data or a program recorded on a removable recording medium 2113 such as a mounted magnetic disk, an optical disk, a magneto-optical disk, or a semiconductor memory, and reads the data or program as an interface 2107 or an external bus 2106. , The bridge 2105, and the RAM 2103 connected via the host bus 2104. The removable recording medium 2113 can also be used as a data recording area similar to the hard disk.

  The connection port 2114 is a port for connecting the external connection device 2115 and has a connection unit such as USB, IEEE1394. The connection port 2114 is connected to the CPU 2101 and the like via the interface 2107, the external bus 2106, the bridge 2105, the host bus 2104, and the like. A communication unit 2116 is connected to the network and executes data communication processing with the outside. The data reading unit 2117 is a scanner, for example, and executes document reading processing. The data output unit 2118 is a printer, for example, and executes document data output processing.

  Note that the hardware configuration of the image processing apparatus shown in FIG. 21 shows one configuration example, and the present embodiment is not limited to the configuration shown in FIG. 21, and the modules described in this 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 the systems shown in FIG. 21 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.).

Various feature extraction modules 180 in the above-described embodiments may be combined, and the technology described in the background art may be adopted as the processing content of each module. For example, as feature quantities, feature quantities extracted by two or more of the peak / teacher data comparison module 942, the peak circumference variance calculation module 944, the peak feature extraction module 1420, and the variance total / height extraction module 1820 are used. You may combine. Further, it may be combined with a feature amount using existing technology (dispersion of spacing between characters, vertical blur, area dispersion, etc.).
Then, the distance measurement module 185 measures the distance in the feature space based on the combination of the feature amounts, and the determination module 190 determines whether the character string in the character string image is a type or handwriting based on the difference in distribution.

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 Disc (registered trademark), magneto-optical disk (MO), flexible disk (FD), magnetic tape, hard disk, read-only memory (ROM), electrically erasable and rewritable read-only memory (EEPROM), flash Includes memory, random access memory (RAM), etc. .
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 110 ... Image reception module 120 ... Character string extraction module 130 ... Thin line processing module 140 ... Rectangle size determination module 150 ... Rectangle division processing module 160 ... Hough conversion processing module 170 ... Threshold processing module 180 ... Feature extraction module 185 ... Distance measurement module 190 ... Determination module 195 ... Output module 910 ... Addition processing module 920 ... Histogram creation processing module 930 ... Peak extraction module 940 ... Peak processing module 942 ... Peak / teacher data comparison module 944 ... Peak peripheral variance calculation module 950 ... Feature quantity storage module 1100 ... Teacher data table 1410 ... Peak extraction module 1420 ... Peak feature extraction module 1430 ... Feature quantity storage module 1810 ... Variance calculation module 1820 ... Variance total / height extraction module 1830 ... Feature quantity storage module

Claims (7)

Fine line conversion means for converting into a thin line information for a character string image including a character string,
Area setting means for setting a rectangular area in the character string image that has been turned into a thin line by the thin line conversion means;
Hough transforming means for performing Hough transform on the rectangular area set by the area setting means;
Feature extraction means for extracting the angle information of the line segment of the Hough transformation processing result by the Hough transformation means and the height information thereof as features,
An image processing apparatus comprising: a determination unit that determines whether a character string in the character string image is a type or a handwriting based on the feature extracted by the feature extraction unit. The feature extraction means includes
A weighting unit for weighting the processing result of each rectangle of the Hough transform by the Hough transforming unit, and an adding unit for adding the weighted processing result excluding those determined not to be line segment information;
A first value extracting means for extracting a first value from the processing result by comparing the processing result by the adding means with a distribution in angle and a predetermined value;
2. The first feature quantity extraction means for extracting, as a feature quantity, a difference between the first value extracted by the first value extraction means and a predetermined value. Image processing apparatus. The feature extraction means includes
A weighting unit for weighting the processing result of each rectangle of the Hough transform by the Hough transforming unit, and an adding unit for adding the weighted processing result excluding those determined not to be line segment information;
A first value extracting means for extracting a first value from the processing result by comparing the processing result by the adding means with a distribution in angle and a predetermined value;
A second feature amount extracting unit for extracting, as a feature amount, a variance of a processing result by the adding unit within a predetermined range with respect to the first value extracted by the first value extracting unit; The image processing apparatus according to claim 1. The feature extraction means includes
Second feature quantity extraction means for extracting, as a feature quantity, a variance of a processing result by the addition means within a predetermined range with respect to the first value extracted by the first value extraction means. The image processing apparatus according to claim 2. The feature extraction means includes
A second value extracting means for extracting a second value from the processing result by comparing the processing result of each rectangle of the Hough transform by the Hough transforming means with a predetermined value;
5. The apparatus according to claim 1, further comprising a third feature amount extraction unit that extracts, as a feature amount, a height and an angle of the second value extracted by the second value extraction unit. An image processing apparatus according to 1. The feature extraction means includes
Dispersion calculating means for calculating the dispersion of the processing results of the respective rectangles of the Hough transform by the Hough transform means within a predetermined range with respect to a predetermined value;
5. The method according to claim 1, further comprising: a fourth feature amount extraction unit that extracts a total and height of the variances of the rectangles calculated by the variance calculation unit as feature amounts. The image processing apparatus described. Computer
Fine line conversion means for converting into a thin line information for a character string image including a character string,
Area setting means for setting a rectangular area in the character string image that has been turned into a thin line by the thin line conversion means;
Hough transforming means for performing Hough transform on the rectangular area set by the area setting means;
Feature extraction means for extracting the angle information of the line segment of the Hough transformation processing result by the Hough transformation means and the height information thereof as features,
An image processing program for causing a character string in the character string image to function as a determination unit that determines whether a character string is a type or a handwriting based on the feature extracted by the feature extraction unit.

Images