JP2019197444A - Image processing program, image processing method, and image processing device - Google Patents

Abstract

To automatically specify an area in the image of a business form correctly in which a date stamp is put.SOLUTION: Provided is an image processing program which causes a computer to execute the process of: extracting a pixel that indicates a portion, out of the pixels in the image of a business form indicating elements of the business form, where the angle formed by angles in the horizontal and vertical directions of the image of the business form in the elements of the business form; grouping a plurality of pixels, out of the extracted pixels, that are adjacent and contiguous as one group; extracting a group, out of the pixel groups, where the lengthwise and breadthwise lengths in the image are greater than or equal to a threshold and where two sets or more of other groups exist that are within the prescribed range of distance; and specifying an area in the image of the business form that includes all of the extracted groups as an area where a date stamp is put.SELECTED DRAWING: Figure 3

Description

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

  As one of image processing apparatuses that perform character recognition on a form image, there is an apparatus that performs processing for recognizing a date included in a date stamp stamped on a form by character recognition. In this type of image processing apparatus, in order to correctly recognize the date included in the date stamp, an area where the date stamp is stamped is cut out from the image of the form and the orientation of the date stamp is corrected.

  When character recognition is performed on the date of the date stamp in the form image, first, the position where the date stamp is stamped in the form image is specified. Patent Document 1 describes that, based on the form ID, a corresponding date stamped area is extracted from the form format data storage unit, and the position where the date stamp is stamped in the form image is specified. Japanese Patent Application Laid-Open No. 2004-228561 describes that an area in which a date stamp is stamped in a form image is specified based on coordinates (xs, xe, ys, ye) indicating the position of a stamp frame in the form. .

JP-A-8-190610 JP-A-6-111066

  However, in recent years, with the increase in private-made forms and the like, the positions of areas for imprinting date stamps in the forms have diversified. For this reason, it takes a lot of labor and time to create the form format data including the position information of the area for imprinting the date stamp for all of the various forms having the area for imprinting the date stamp.

  Furthermore, since the person stamps the date stamp on the form, a part of the stamped date stamp may protrude outside the area to be stamped. In this way, when a part of the date protrudes outside the area to be stamped, the date stamp in the area cut out from the form image based on the position of the area to be stamped is partially missing. This makes it difficult to correctly recognize the date stamp direction and date. If the date of the date stamp cannot be correctly recognized, the operator needs to confirm the date stamp stamped on the original form (paper medium) and enter the correct date, which increases the labor of the operator. .

  The present invention has been made in view of the above-described problems, and an object of the present invention is to provide a technique capable of automatically and correctly specifying a region where a date stamp is imprinted in a form image.

  The image processing program according to the first aspect of the present invention includes an angle formed between each of a horizontal direction and a vertical direction of the form image in the form element among pixels indicating the form element in the form image. Pixels indicating a portion where the angle is equal to or greater than a predetermined angle, the extracted pixels are grouped as a group of a plurality of adjacent continuous pixels, and the vertical and horizontal directions in the image are out of the group of pixels. A group in which two or more other groups whose length is equal to or greater than the threshold and whose distance is within a predetermined range is extracted, and an area including all of the extracted groups in the form image is represented by a date stamp. This is an image processing program for causing a computer to execute processing for making an area marked with.

  In the image processing method according to the second aspect of the present invention, among the pixels indicating the element of the form in the form image, the computer processes each of the horizontal and vertical directions of the form image in the form element. A pixel indicating a portion where the angle formed by is equal to or greater than a predetermined angle is extracted, and the extracted pixels are grouped as a group of a plurality of adjacent pixels that are adjacent to each other. Extract a group in which two or more other groups having a vertical and horizontal length equal to or greater than a threshold and a distance within a predetermined range exist, and include an area including all of the extracted groups in the form image This is an image processing method for executing a process for setting a date stamped area.

  An image processing apparatus according to a third aspect of the present invention provides an angle formed between each of a horizontal direction and a vertical direction of the form image in the form element among pixels indicating the form element in the form image. After extracting pixels indicating a portion where is equal to or greater than a predetermined angle, the extracted pixels are grouped as a group of a plurality of adjacent and continuous pixels, and then within the image of the group of pixels Extract a group in which two or more other groups having a vertical and horizontal length equal to or greater than a threshold and a distance within a predetermined range exist, and include an area including all of the extracted groups in the form image It is an image processing apparatus provided with the analysis part made into the area | region where the date stamp was impressed.

  According to the present invention, it is possible to automatically and correctly specify a region where a date stamp is imprinted in a form image.

It is a figure which shows the structure of an image processing system. It is a figure which shows the functional structure of the image processing apparatus which concerns on one Embodiment. It is a flowchart explaining the image processing which concerns on one Embodiment. It is a flowchart explaining the content of an inclination extraction process. It is a flowchart explaining the content of a noise removal process. It is a flowchart explaining the content of a grouping process. It is a flowchart explaining the content of an erecting correction process. It is a figure which shows the example of a form. It is a figure which shows the example of the date stamp stamped on the form. It is a figure which shows another example of the date stamp stamped on the form. It is a figure explaining the calculation method of an inclination. It is a figure which shows the calculation result of an inclination. It is a figure which shows the extraction result of inclination. It is FIG. (1) explaining the method to group a pixel. It is FIG. (2) explaining the method to group a pixel. It is a figure explaining the removal method of noise. It is a figure explaining the method of calculating the inclination of a date stamp. It is a figure explaining a mode that a date stamp is rectified upright. It is a figure which shows the hardware constitutions of a computer.

FIG. 1 is a diagram illustrating a configuration of an image processing system.
The image processing system 1 in FIG. 1 is a system that automatically performs processing for recognizing a date included in a date stamp stamped on a form by character recognition. The image processing system 1 includes an image processing device 2, a scanner device 3, an input device 4, a display device 5, and a server device 6.

  The image processing apparatus 2 identifies an area where a date stamp is impressed in a form image, and recognizes a date included in the date stamp by character recognition. For example, the image processing apparatus 2 acquires a form image from the scanner apparatus 3. The scanner device 3 is a device that converts a paper medium such as a form into digital data (image).

  The input device 4 is a device that inputs various types of information to the image processing device 2. The input device 4 is, for example, a keyboard device. The display device 5 is a device that displays various types of information such as the contents of processing performed by the image processing device 2 and the results of the processing, information input by the input device 4, images of forms, and the like. The display device 5 is, for example, a liquid crystal display.

  The server device 6 is a device that holds a character recognition dictionary that the image processing device 2 refers to when recognizing characters in an image. The server device 6 is communicably connected to the image processing device 2 via the network 7.

  After acquiring the form image from the scanner device 3, the image processing apparatus 2 first specifies the position (area) where the date stamp is impressed in the image. Thereafter, the image processing apparatus 2 calculates the inclination of the date stamp in the image, and corrects the image so that the orientation of the date stamp is upright when the inclination is equal to or greater than a predetermined angle. Then, the image processing apparatus 2 performs character recognition in a state where the direction of the date stamp in the image is upright, and recognizes the date included in the date stamp. The image processing device 2 causes the display device 5 to display the form image and the recognized date. Note that the operator of the image processing system 1 operates the input device 4 to set the date when, for example, the date recognition result displayed on the display device 5 is different from the date stamped on the form. It is possible to correct.

FIG. 2 is a diagram illustrating a functional configuration of the image processing apparatus according to the embodiment.
As shown in FIG. 2, the image processing apparatus 2 according to the present embodiment includes a scanner control unit 210, an analysis unit 220, a recognition processing unit 230, a display control unit 240, and a main control unit 250. The image processing apparatus 2 includes a storage unit (not shown).

The scanner control unit 210 controls processing for acquiring a form image from the scanner device 3.
The analysis unit 220 analyzes the position and inclination of the date stamp in the form image, and corrects the image so that the orientation of the date stamp in the image is upright. The analysis unit 220 includes a conversion unit 221, a gradient extraction unit 222, a noise removal unit 223, a cutout unit 224, and a correction unit 225.

  When the form image is a color image, the conversion unit 221 converts the color image into a grayscale image.

  The gradient extraction unit 222 extracts a portion of the form element (line, character, figure, date stamp, etc.) in the form image where the angle between the horizontal direction and the vertical direction of the image is a predetermined angle or more. To do. For example, the gradient extraction unit 222 calculates the gradient for each pixel indicating any part of the form element existing in the image based on the density value of the pixels around the pixel and a filter described later. Then, a pixel whose calculated inclination is equal to or greater than a threshold is extracted. The slope is a value corresponding to the size of the angle formed between the stretching direction of the pixel indicated by the slope and the surrounding pixels of the form elements and the horizontal and vertical directions of the image. Among the elements of the form, the inclination when the extending direction of the pixel indicated by the pixels and the surrounding pixels is substantially parallel to either the image horizontal direction or the image vertical direction is the pixel for calculating the inclination and This is a value smaller than the inclination when the extending direction of the portion indicated by the surrounding pixels is oblique with respect to the horizontal direction and the vertical direction of the image.

  The noise removing unit 223 excludes pixels indicating an element different from the date stamp among pixels whose gradient extracted by the gradient extracting unit 222 is equal to or greater than a threshold from the pixel candidates indicating a part of the date stamp as noise. To do. The noise removing unit 223 identifies pixels indicating elements different from the date stamp based on the features of the date stamp's external shape and dimensions in the image. For example, the noise removing unit 223 groups pixels having the gradient extracted by the gradient extracting unit 222 with a threshold value equal to or greater than a threshold for each of a plurality of adjacent pixels, and the vertical and horizontal dimensions are included in a group smaller than the threshold. Pixels are excluded from pixel candidates that show part of a date stamp as noise. In addition, for example, the noise removing unit 223 selects pixels included in a group in which two or more other groups whose distances are within a predetermined range from among groups whose vertical and horizontal dimensions are equal to or larger than a threshold value, Extracted as a pixel indicating a part. In other words, the noise removal unit 223 is included in a group in which only one set of other groups whose distances are within a predetermined range and a group in which no set exists, among the groups whose vertical and horizontal dimensions are equal to or greater than the threshold value. Are excluded from candidate pixels that indicate part of the date stamp as noise.

  Based on the position in the image of the pixel included in the group that is not excluded by the noise removal unit 223 among the pixels whose gradient extracted by the gradient extraction unit 222 is greater than or equal to the threshold, the cutout unit 224 The area with the date stamp inside is cut out. For example, the cutout unit 224 identifies a rectangular region including all of the groups that are not excluded by the noise removal unit 223 in the image as a region where the date stamp is stamped, and cuts out the region.

  The correction unit 225 rotationally corrects the cutout region so that the orientation of the date stamp is upright when the inclination of the date stamp in the cutout region with respect to the horizontal direction of the image is equal to or greater than the threshold. The correction unit 225 estimates, for example, the inclination of the first straight line portion located above the date and the second straight line portion located below the date in the date stamp with respect to the image horizontal direction by Hough transform, Based on the estimation result, the position (coordinates) of the pixel in the cut out region is converted so that the direction of the date stamp is established.

  Based on the analysis result of the analysis unit 220, the recognition processing unit 230 recognizes the date included in the area where the date stamp cut out from the form image is stamped by character recognition. The recognition processing unit 230 refers to, for example, the character recognition dictionary 610 held by the server device 6, and between the first straight line part and the second straight line part in the region (partial image) where the date stamp is stamped. Recognize numbers and symbols lined up in the area. Note that the character recognition dictionary 610 referred to by the recognition processing unit 230 may be stored in, for example, a storage unit (not shown) included in the image processing apparatus 2.

  The display control unit 240 controls the display of the display device 5. For example, the display control unit 240 generates screen data including a form image and a date recognition result of the date stamp, and causes the display device 5 to display the screen data.

  The main control unit 250 controls a series of processing performed in cooperation by the scanner control unit 210, the analysis unit 220, the recognition processing unit 230, and the display control unit 240.

  As described above, the image processing apparatus 2 according to the present embodiment is based on the characteristics of the inclination and the external shape and size of the date stamp corresponding to the angles formed by the horizontal and vertical directions of the image of the form. The analysis unit 220 for specifying the position where the date stamp in the image is stamped. For this reason, in the image processing apparatus 2 of this embodiment, it is possible to specify an area including the entire date stamp regardless of the position of the date stamp stamped on the form. In addition, the analysis unit 220 in the image processing apparatus 2 according to the present embodiment further calculates (estimates) the inclination of the date stamp in the image of the form, and corrects the date stamp to be erect. For this reason, in the image processing apparatus 2 according to the present embodiment, it is possible to perform character recognition in a state where the direction of the date included in the date stamp is upright.

FIG. 3 is a flowchart illustrating image processing according to an embodiment.
When image processing is started, the image processing apparatus 2 according to the present embodiment first acquires an image of a form as shown in FIG. 3 (step S1). The process of step S1 is performed by the scanner control unit 210 of the image processing apparatus 2. For example, the scanner control unit 210 acquires an RGB color image of a form from the scanner device 3.

  Next, the image processing apparatus 2 converts the acquired form image into a grayscale image (step S2). The processing in step S2 is performed by the conversion unit 221 in the analysis unit 220 of the image processing apparatus 2. The conversion unit 221 converts the RGB color image into a grayscale image having a predetermined gradation according to a known conversion method. For example, based on the RGB luminance value of each pixel in the RGB color image, the conversion unit 221 converts the background density (gradation) in the form that is a paper medium into the density of elements such as ruled lines, characters, and date stamps in the form ( The RGB color image is converted into a gray scale image so as to be higher than (gradation). If the form image acquired in step S1 is a grayscale image or a binarized monochrome image, the process in step S2 may be omitted.

  Next, the image processing apparatus 2 performs an inclination extraction process (step S3) and a noise removal process (step S4) as a process for acquiring information used for specifying the position of the date stamp in the form image. The slope extraction process and the noise removal process are an example of a process of extracting a part of an element in an image where the angle between the image horizontal direction and the image vertical direction is greater than or equal to a threshold value as a date stamp candidate.

  The gradient extraction process in step S3 is performed by the gradient extraction unit 222 in the analysis unit 220 of the image processing apparatus 2. The inclination extraction unit 222 calculates an inclination corresponding to an angle formed between the horizontal direction and the vertical direction of the image for each pixel whose density corresponding to a line, character, date stamp, or the like in the image is a threshold value or less. Pixels for which the slope is equal to or greater than the threshold are extracted. Here, the slope is a value calculated by the density of a pixel adjacent to the target pixel for calculating the slope and a predetermined filter. For example, in the filter, the slope when the pixel of interest is one of a plurality of pixels having a density equal to or less than a threshold value that is continuous in the oblique direction in the image is continuously horizontal or vertical in the image. This is a set of coefficients set so as to have a value larger than the inclination when any one of the plurality of pixels whose density is equal to or less than the threshold value is the target pixel.

  The noise removal processing in step S4 is performed by the noise removal unit 223 in the analysis unit 220 of the image processing apparatus 2. The noise removing unit 223 uses pixels corresponding to elements (ruled lines, characters, etc.) other than the elements included in the date stamp in the image, which are included in the pixels whose gradient extracted in step S3 is equal to or greater than the threshold value, as noise. It excludes from the pixel candidate which shows the element of a mark. For example, the noise removing unit 223 groups pixels having an inclination equal to or greater than a threshold value, and includes pixels included in a group that does not satisfy the conditions set based on the shape and size characteristics of the date stamp in the image. It excludes from the pixel candidate which shows the element of a date stamp.

  When the processing of steps S3 and S4 is completed, the image processing apparatus 2 next specifies the position of the date stamp and cuts out the area where the date stamp is stamped (step S5). The processing in step S5 is performed by the cutout unit 224 in the analysis unit 220 of the image processing apparatus 2. The cutout unit 224 calculates a rectangular area including all the candidate pixels (groups) indicating the date stamp elements obtained by the processes in steps S3 and S4, and cuts out the rectangular area from the form image.

  When the process of step S5 is completed, the image processing apparatus 2 next performs an erecting correction process (step S6) for correcting the date mark included in the cut-out image to be erect. The upright correction process in step S6 is performed by the correction unit 225 in the analysis unit 220 of the image processing apparatus 2. The correction unit 225 calculates the inclination of the date stamp in the cut-out image, and rotates the cut-out image by an angle corresponding to the inclination angle when the inclination is equal to or greater than the threshold value. The correction unit 225 calculates the inclination angle θ of the date stamp according to a known method, and then rotates the image by −θ according to the known method. For example, the correction unit 225 extracts the first straight line portion located above the date and the second straight line portion located below the date in the date stamp by the Hough transform, and the straight line portion from the horizontal direction is extracted. The inclination angle θ is calculated. Thereafter, the correction unit 225 rotates the entire cut image by −θ. As a result, the extending direction of the first straight line portion located above the date and the second straight line portion located below the date in the date stamp in the cutout image substantially coincides with the horizontal direction, and the date direction is correct. It becomes a standing state.

  Next, the image processing apparatus 2 recognizes the date included in the date stamp by character recognition (step S7). The recognition processing unit 230 of the image processing apparatus 2 performs the process in step S7. The recognition processing unit 230 extracts a region indicating the date in the image in which the direction of the date mark is in an upright state, performs character recognition on the region, and recognizes the date. For example, the recognition processing unit 230 is surrounded by a first straight line portion located above the date extracted in step S6, a second straight line portion located below the date, and a line indicating the outline of the date stamp. This area is extracted as an area indicating the date. Further, the recognition processing unit 230 recognizes numbers and symbols arranged in the area indicating the date according to a known recognition method. At this time, the recognition processing unit 230 refers to, for example, the character recognition dictionary 610 of the server device 6 and recognizes numbers and symbols arranged in the area indicating the date. Note that the character recognition dictionary 610 referred to by the recognition processing unit 230 may be stored in a storage unit (not shown) provided in the image processing apparatus 2.

  When the process of step S7 is completed, the image processing apparatus 2 outputs a recognition result (step S8), and ends a series of image processing. The display control unit 240 of the image processing apparatus 2 performs the process in step S8. For example, the display control unit 240 creates screen data including the form image acquired in step S1, the date stamp image corrected in step S6, and the date recognition result, and causes the display device 5 to display the screen data.

  As described above, the image processing apparatus 2 according to the present embodiment performs the gradient extraction process and the noise removal process, and the angle formed between the image horizontal direction and the image vertical direction in the date stamp in the image is greater than or equal to the threshold value. Extract the part.

FIG. 4 is a flowchart for explaining the content of the gradient extraction process.
As described above, the gradient extraction process (step S3) is performed by the gradient extraction unit 222 in the analysis unit 220 of the image processing apparatus 2. The gradient extraction unit 222 performs loop processing (steps S301 to S306) as shown in FIG. 4 as the gradient extraction processing. The loop processing is performed for each pixel in the form image.

  At the start of the loop processing (step S301), the gradient extraction unit 222 selects a pixel to be processed from among pixels in the image. For example, the gradient extraction unit 222 selects pixels to be processed by raster scanning. On the other hand, at the end of the loop processing (step S306), the gradient extraction unit 222 determines whether or not processing has been performed on all the pixels in the form image. If there is a pixel that has not been processed, the gradient extraction unit 222 continues the loop processing (steps S301 to S306). When processing is performed for all pixels, the gradient extraction unit 222 ends the loop processing (steps S301 to S306) and ends the gradient extraction processing.

  After selecting the pixel to be processed in step S301, the gradient extraction unit 222 determines whether or not the density of the selected pixel is equal to or lower than the threshold value TH1 (step S302). The threshold value TH1 is set to a value between the density of the background in the grayscale image of the form and the density of the form elements such as lines, characters, and date stamps. In step S2, when the form image is converted into a grayscale image in which the background density of the form is higher than the density of the elements (lines, characters, date stamps, etc.), pixels whose density is higher than the threshold TH1 This is a pixel indicating the background. For this reason, when the density of the selected pixel is higher than the threshold value TH1 (step S302; YES), the gradient extraction unit 222 omits the processes of steps S303 to S305, and loops at the end of the loop process (step S306). It is determined whether or not to continue the process.

On the other hand, a pixel whose density of the selected pixel is equal to or lower than the threshold value TH1 is a pixel indicating a form element including lines, characters, date stamps, and the like in the form. For this reason, when the density of the selected pixel is equal to or lower than the threshold value TH1 (step S302; YES), the gradient extraction unit 222 next calculates the slope of the corresponding pixel based on the density of the pixels around the corresponding pixel and the filter. The degree is calculated (step S303). In step S303, the gradient extraction unit 222 calculates the gradient S _{u, v} of the pixel P _{u, v} by the following mathematical formula (1), for example.

S _{u, v} = a 1 · D _{u-1, v-1} + a 2 · D _{u, v-1} + a 3 · D _{u + 1, v-1}
+ A4 · D _{u-1, v} + a5 · D _{u + 1, v}
+ A6.Du _{-1, v + 1} + a7.Du _{, v + 1} + a8.Du _{+ 1, v + 1} (1)

The pixel P _{u, v} has coordinates (u, v when the pixel at the upper left corner in the image of the form is the origin (u = 0, v = 0), the right direction is + u direction, and the lower direction is + v direction. It is a pixel specified in v). Also, D _{u−1, v−1} , D _{u, v−1} , D _{u + 1, v−1} , D _{u−1, v} , D _{u + 1, v} , D _u−1 in equation (1) _{, v + 1} , D _{u, v + 1} , and D _{u + 1, v + 1} are pixels P _{u−1, v−1} , P _{u, v−1} , P _{u + 1, v−1, respectively.} , P _{u−1, v} , P _{u + 1, v} , P _{u−1, v + 1} , P _{u, v + 1} , and P _{u + 1, v + 1} . In addition, a1 to a8 in Expression (1) are filter coefficients, respectively. In this embodiment, the filter coefficients a1 to a8 are, for example, a1 = a8 = 1, a2 = a4 = a5 = a7 = 0, and a3 = a6 = -1.

After calculating the slope of the selected pixel, the slope extraction unit 222 next determines whether or not the calculated absolute value of the slope is equal to or greater than the threshold value TH2 (step S304). The threshold value TH2 is an average inclination of a pixel indicating a portion extending obliquely with respect to the horizontal direction and the vertical direction in a form element (line, character, date stamp, etc.) and a portion extending in the horizontal direction and the vertical direction. Is set to a value between the average inclination of the pixels indicating The slope S _{u, v} calculated by Equation (1) is a portion where the selected pixel P _{u, v} extends obliquely with respect to the horizontal and vertical directions in the form elements (lines, characters, date stamps, etc.). The absolute value in the case of a pixel that indicates is larger than the absolute value in the case of a pixel that indicates a portion extending in the horizontal direction and the vertical direction. That is, the pixel whose absolute value of the calculated inclination is smaller than the threshold value TH2 is a pixel indicating a portion extending in the horizontal direction and the vertical direction in the form element (line, character, date stamp, etc.). Therefore, if the slope is smaller than the threshold value TH2 (step S304; NO), the slope extraction unit 222 omits the process of step S305 and continues the loop process at the end of the loop process (step S306). Determine whether or not. That is, the gradient extraction unit 222 indicates a portion extending in the horizontal direction and the vertical direction in the form elements (lines, characters, date stamps, etc.) among the pixels indicating the form elements whose density is equal to or less than the threshold value TH1. The pixel is excluded from the pixels from which the slope is extracted.

  On the other hand, a pixel whose slope is equal to or greater than the threshold TH2 is a pixel that indicates a portion extending in an oblique direction with respect to a horizontal direction and a vertical direction in a form element (line, character, date stamp, etc.). For this reason, when the absolute value of the gradient is equal to or greater than the threshold value TH2 (step S304; YES), the gradient extraction unit 222 holds the gradient of the corresponding pixel (step S305). Thereafter, the gradient extraction unit 222 determines whether or not to continue the loop processing at the end of the loop processing (step S306).

  As described above, in the gradient extraction process, only the gradient of the pixel indicating the portion extending in the oblique direction with respect to the horizontal direction and the vertical direction among the elements (lines, characters, date stamps, etc.) of the form in the form image. To extract. That is, by performing the gradient extraction process, pixels indicating a portion extending in an oblique direction with respect to the horizontal direction and the vertical direction in the form elements (lines, characters, date stamps, etc.) in the form image are specified. The The ruled lines in the form image are generally a straight line extending in the horizontal direction of the image and a straight line extending in the vertical direction of the image. On the other hand, as will be described later, the date stamp includes a circular or polygonal outline, and there are portions extending in an oblique direction with respect to each of the image horizontal direction and the image vertical direction. For this reason, it is possible to narrow down the position (area) where the date stamp is stamped in the image of the form by performing the gradient extraction process.

  In some forms, corners of the frame line are arc-shaped, or characters including a portion extending in an oblique direction with respect to each of the image horizontal direction and the image vertical direction are described. When the gradient extraction process is performed on this type of form, the gradient of the pixel indicating the arcuate corner or the pixel indicating a part of the character is equal to or greater than the threshold value TH2, and the gradient of the corresponding pixel is extracted. For this reason, in the image processing according to the present embodiment, after the gradient extraction process, the noise removal process (step S4) is performed, and the shape and size characteristics of the date stamp outline of the pixels from which the gradient is extracted are obtained. Pixels that do not satisfy the conditions shown are excluded from image candidates that show date stamps as noise.

  FIG. 5 is a flowchart for explaining the contents of the noise removal processing. FIG. 6 is a flowchart for explaining the contents of the grouping process.

  As described above, the noise removal processing (step S4) is performed by the noise removal unit 223 in the analysis unit 220 of the image processing apparatus 2. The noise removal unit 223 performs processing according to the flowchart of FIG. 5 as noise removal processing.

  First, the noise removing unit 223 performs a grouping process (step S401) for grouping pixels from which the slope has been extracted in the slope extraction process. For example, the noise removing unit 223 performs processing according to the flowchart of FIG. 6 as the grouping processing.

  In the grouping process, the noise removal unit 223 first initializes the group number GN to be assigned to the pixel (step S401a). In step S401a, the noise removing unit 223 sets GN = 1.

  Next, the noise removal unit 223 performs a loop process (Steps S401b to S401j). The loop processing is performed for each pixel in the form image.

  At the start end of the loop processing (step S401b), the noise removal unit 223 selects a pixel to be processed from among the pixels in the image. For example, the noise removal unit 223 selects a pixel to be processed by raster scanning. On the other hand, at the end of the loop processing (step S401j), the noise removing unit 223 determines whether or not processing has been performed on all pixels in the form image. If there is a pixel that has not been processed, the noise removing unit 223 continues the loop processing (steps S401b to S401j). When the process is performed on all the pixels, the noise removing unit 223 ends the loop process (Steps S401b to S401j) and performs a process of updating the group number (Step S401k).

  After selecting a pixel to be processed in step S401b, the noise removal unit 223 determines whether or not the inclination of the pixel is held (step S401c). When the corresponding inclination is not held (step S401c; NO), the noise removing unit 223 omits the processing after step S401d and determines whether or not to continue the loop processing at the end of the loop processing (step S401j). judge.

  On the other hand, when the inclination of the corresponding pixel is held (step S401c; YES), the noise removal unit 223 next assigns group numbers to the upper left, upper, upper right, and left pixels of the corresponding pixel. It is determined whether or not (step S401d). When there is a pixel that is not assigned a group number among the upper left, upper, upper right, and left pixels of the corresponding pixel (step S401d; NO), the noise removal unit 223 determines that the corresponding pixel (the pixel selected in step S401b) Is assigned a group number GN (step S401h). After the process of step S401h, the noise removing unit 223 updates the group number GN to GN + 1 (step S401i), and determines whether or not to continue the loop process at the end of the loop process (step S401j).

  On the other hand, when the group number is assigned to all of the upper left, upper, upper right, and left pixels of the corresponding pixel (step S401d; YES), the noise removing unit 223 is then assigned to each pixel. The smallest group number is assigned to the corresponding pixel (the pixel selected in step S401b) (step S401e).

  When the process of step S401e is completed, the noise removing unit 223 next determines whether or not the group numbers of the upper left, upper, upper right, and left pixels of the corresponding pixel are all the same (step S401f). When all the group numbers are the same (step S401d; YES), the noise removing unit 223 omits the process of step S401g and determines whether to continue the loop process at the end of the loop process (step S401j). .

  On the other hand, when there are two or more group numbers assigned to the upper left, upper, upper right, and left pixels of the corresponding pixel (step S401f; NO), the noise removing unit 223 performs the upper left, upper, upper right, and left The smallest group number among the group numbers assigned to the pixels is registered in the replacement table (step S401g). In the process of step S401g, the noise removing unit 223 associates a non-minimum group number and a minimum group number among two or more group numbers assigned to the upper left, upper, upper right, and left pixels. . When the process of step S401g is completed, the noise removing unit 223 determines whether or not to continue the loop process at the end of the loop process (step S401j). When continuing the loop processing, the noise removing unit 223 selects a pixel to be subjected to the next loop processing at the start end of the loop processing (step S401b). When the loop processing is performed on all the pixels, the noise removing unit 223 ends the loop processing, and then updates the group number based on the replacement table (step S401k). In the process of step S401k, the noise removing unit 223 refers to the replacement table, and among the group numbers assigned to the pixels, the group number of the pixel associated with the minimum group number by the process of step S401g is Replace with the lowest group number. For example, in the replacement table, when the minimum group number “1” is associated with the pixel assigned with the group number “3”, the noise removing unit 223 changes the group number of the corresponding pixel from “3” to “3”. Update to “1”.

  When the process of step S401k is completed, the noise removing unit 223 ends the grouping process. That is, when the process of step S401k is completed, the noise removing unit 223 ends the grouping process (step S401) of FIG. 5, and then performs a first loop process (steps S402 to S405). The first loop process is performed for each group determined by the grouping process.

  At the start end of the first loop process (step S402), the noise removing unit 223 selects a group to be processed from the groups determined by the grouping process. For example, when the group number is an integer equal to or greater than 1 as in the grouping process of FIG. 6, the noise removal unit 223 selects in order from the group with the smallest group number. On the other hand, at the end of the first loop processing (step S405), the noise removal unit 223 determines whether or not processing has been performed for all groups. If there is a group that has not been processed, the noise removal unit 223 continues the first loop processing (steps S402 to S405). When the process is performed for all the groups, the noise removing unit 223 ends the first loop process (Steps S402 to S405) and performs the second loop process (Steps S406 to S409).

  After selecting the group to be processed in step S402, the noise removal unit 223 determines whether the vertical and horizontal lengths of the group are equal to or greater than the threshold value TH3 (step S403). The threshold value TH3 is based on the distribution of a plurality of pixels grouped into one group in the grouping process (step S401) among the pixels indicating the outline portions of various date stamps in the form image. It is set to a value smaller than the length of. Further, the threshold value TH3 is set to a value larger than a value that can be considered as the length in the vertical and horizontal directions of the arcuate corners of the above-described form frame line and the diagonally extending portion of the characters. That is, it is very likely that a pixel included in a group whose vertical and horizontal lengths are shorter than the threshold value TH3 is not a pixel indicating a date stamp element (an outline or a linear portion positioned above and below the date). For this reason, when the vertical and horizontal lengths of the corresponding group are shorter than the threshold value TH3 (step S403; NO), the noise removal unit 223 excludes the corresponding group from the group candidates indicating date stamp elements as invalid groups ( Step S404). That is, in step S404, the noise removing unit 223 removes, as noise, pixels included in the group whose length and width of the portion extending in the oblique direction are shorter than the threshold value TH3 from the candidate pixels indicating the date stamp element. To do. When the process of step S404 is completed, the noise removing unit 223 determines whether or not to continue the first loop process at the end of the first loop process (step S405).

  On the other hand, when the vertical and horizontal lengths of the group are greater than or equal to the threshold value TH3 (step S403; YES), the noise removing unit 223 omits the process of step S404 and at the end of the first loop process (step S405). It is determined whether or not to continue the first loop process.

  When there is a group that is not performing the first loop process, the noise removing unit 223 continues the first loop process (steps S402 to S405). When the process is performed for all the groups, the noise removing unit 223 ends the first loop process (Steps S402 to S405) and performs the second loop process (Steps S406 to S409). The second loop process is performed for each group determined by the grouping process and not excluded (removed) in the first loop process.

  At the start end of the second loop process (step S406), the noise removal unit 223 selects a group to be processed from among the groups that are not excluded in the first loop process (effective groups). For example, when the group number is an integer equal to or greater than 1 as in the grouping process of FIG. 6, the noise removal unit 223 selects in order from the group with the smallest group number. On the other hand, at the end of the second loop processing (step S409), the noise removal unit 223 determines whether or not processing has been performed for all the groups to be processed. If there is a group that has not been processed, the noise removal unit 223 continues the second loop processing (steps S406 to S409). When the process is performed for all groups, the noise removal unit 223 ends the second loop process (steps S406 to S409) and ends the noise removal process.

  After selecting a group to be processed in step S406, the noise removal unit 223 determines whether there are two or more effective groups around the group (step S407). An effective group is a group that is not excluded in the first loop process among the groups determined by the grouping process. In the process of step S407, for example, the noise removing unit 223 determines whether or not there is a pixel included in an effective group within a predetermined range (within a region) centered on a predetermined position in the group currently selected for processing. And whether or not there are two or more other effective groups that partially or wholly are included is determined. At this time, the search range of the pixels included in the effective group is set to a range slightly larger than the maximum value of the outer dimensions of the date stamp in the form image, for example.

  Many of the outer shapes of date stamps are circular or polygonal, and the outlines of these date stamps are vertically symmetric and symmetric and have high symmetry. For this reason, in the grouping process (step S401), two or more groups of pixels indicating the outline of the date stamp are extracted, and the distance between the plurality of groups indicating the outline of the date stamp depends on the external dimensions of the date stamp. Distance. Therefore, if there is only one other effective group within the predetermined range centered on the group currently selected for processing, or if there is no other group, the pixels included in the selected group Is very likely to be a pixel indicating an element other than the date stamp element. For this reason, when the effective group which exists around the group currently selected as the object of processing is one set, or when there is no effective group (Step S407; NO), noise removal part 223 is selected. Are excluded from the group candidates indicating date stamp elements as invalid groups (step S408). That is, in step S408, the noise removing unit 223 removes, as noise, a pixel included in a group in which two or more other effective groups are not present from the candidate pixels indicating the date stamp element. When the process of step S408 is completed, the noise removing unit 223 determines whether or not to continue the second loop process at the end of the second loop process (step S409).

  On the other hand, when two or more other effective groups exist in the surroundings (step S407; YES), the noise removing unit 223 omits the process of step S408 and ends the second loop process (step S409). ) To determine whether or not to continue the second loop process.

  When there is a group that is not performing the second loop process, the noise removing unit 223 continues the second loop process (steps S406 to S409). When the process is performed for all groups, the noise removal unit 223 ends the second loop process (steps S406 to S409) and ends the noise removal process.

  As described above, in the image processing apparatus 2 of the present embodiment, most of the date stamps in the image of the form have a plurality of portions extending in an oblique direction with respect to the horizontal direction and the vertical direction of the image in the outline portion. By utilizing this, a pixel indicating the outline of the date stamp in the form image is extracted. Further, in the image processing apparatus 2 of the present embodiment, the pixel indicating the outline of the date stamp is specified using the slope calculated based on the density of the pixel in the grayscale image. That is, in the image processing apparatus 2 of the present embodiment, the position of the date stamp in the form image is specified by a method different from the pattern matching based on the outer shape and color of the date stamp. For this reason, in the image processing apparatus 2 of the present embodiment, the position of the date stamp can be easily specified from the image of the form on which the date stamp having a polygon (for example, an octagon) is imprinted. . Further, in the image processing apparatus 2 of the present embodiment, the position of the date stamp can be specified regardless of the color of the date stamp. Furthermore, in the image processing apparatus 2 according to the present embodiment, other elements of the form exist in the area where the date stamp is impressed, as in the case where the position of the date mark protrudes from the area designated on the form. Even in this case, the position of the date stamp can be specified. Therefore, in the image processing apparatus 2 according to the present embodiment, it is possible to automatically specify the position of the date stamp from the image of the form without the operation by the operator.

  Moreover, the image processing performed by the image processing apparatus 2 according to the present embodiment is an erecting correction process (step S6) for correcting the direction of the date mark in the form image to be erect as shown in the flowchart of FIG. including.

FIG. 7 is a flowchart for explaining the contents of the erecting correction process.
As described above, the upright correction process (step S6) is performed by the correction unit 225 in the analysis unit 220 of the image processing apparatus 2. The correction unit 225 performs processing according to the flowchart of FIG. 7 as the upright correction processing.

  First, the correction unit 225 binarizes the date stamp area cut out from the form image (step S601). The date stamp area is an area (partial image) in which the date stamp cut out from the form image is stamped in the process of step S5 in the flowchart of FIG. Note that the form image from which the date stamp area is cut out is not limited to the color image acquired in step S1, but may be a grayscale image obtained by the conversion process in step S2. In the processing of step S601, the correction unit 225 sets the pixel indicating the background (paper color) among the pixels in the date stamp area to “0”, and indicates the form element (date stamp, line, character, etc.) Is “1”.

  Next, the correction unit 225 performs a first loop process (steps S602 to S608). The first loop process is performed for each pixel in the date stamp area.

  At the start end of the first loop process (step S602), the correction unit 225 selects a pixel to be processed from among the pixels in the date stamp area. For example, the correction unit 225 selects a pixel to be processed by raster scanning. On the other hand, at the end of the first loop process (step S608), the correction unit 225 determines whether or not the process has been performed on all the pixels in the date stamp area. If there is a pixel that has not been processed, the correction unit 225 continues the first loop processing (steps S602 to S608). When processing has been performed for all pixels, the correction unit 225 ends the first loop processing (steps S602 to S608), calculates the inclination angle of the date stamp, and rotates the date stamp area (steps S609 to S609). S611) is performed.

  After selecting the pixel to be processed in step S602, the correction unit 225 determines whether or not the value assigned to the pixel in the binarization process is “0” (step S603). A pixel having a value of “0” is a pixel indicating the background (paper color) of the form. For this reason, when the value of the pixel is “0” (step S603; YES), the correction unit 225 omits the second loop processing (steps S604 to S607), and ends the first loop processing (step In S608), it is determined whether or not to continue the first loop process.

On the other hand, when the value of the pixel is other than “0” (step S603; NO), the correction unit 225 next performs a second loop process (steps S604 to S607). The second loop process is performed for each angle θ included in the set of a plurality of angles θ selected within the range of 0 ≦ θ <2π. For example, the set of a plurality of angles θ includes N kinds of angles θ _n = (2π / 100) · n {n = 0, 1, 2,..., N−1}. N is an arbitrary integer, for example, N = 100.

  At the start end of the second loop process (step S604), the correction unit 225 selects the angle θ to be processed from the set of the plurality of angles θ. For example, the correction unit 225 selects in order from the angle θ having the smallest value. On the other hand, at the end of the second loop process (step S607), the correction unit 225 determines whether or not the process has been performed at all angles θ. When there is an angle θ that has not been processed, the correction unit 225 continues the second loop processing (steps S604 to S607). When the processing is performed at all angles θ, the correction unit 225 ends the second loop processing (steps S604 to S607), and performs the first loop processing at the end of the first loop processing (step S608). Determine whether to continue.

  After selecting the angle θ to be processed in step S604, the correction unit 225 calculates ρ = x · cos θ + y · sin θ with respect to the coordinates (x, y) of the pixel currently being processed (step S605). ), (Θ, ρ) is plotted in the θ-ρ coordinate system (step S606). When the processes of steps S605 and S606 are completed, the correction unit 225 determines whether or not to continue the second loop process at the end of the second loop process (step S607). If there is an angle θ for which the processes of steps S605 and S606 are not performed, the correction unit 225 continues the second loop process. When the processes of steps S605 and S606 are performed at all angles θ, the correction unit 225 ends the second loop process, and performs the first loop process at the end of the first loop process (step S608). Determine whether to continue. If there is a pixel that has not been processed, the correction unit 225 continues the first loop processing (steps S602 to S608). When processing has been performed for all pixels, the correction unit 225 ends the first loop processing (steps S602 to S608), calculates the inclination angle of the date stamp, and rotates the date stamp area (steps S609 to S609). S611) is performed.

  When the first loop processing is finished, the correction unit 225 is then plotted from the points (θ, ρ) plotted in the θ-ρ coordinate system based on the result of the first loop processing. The point (θ1, ρ1) and the point (θ2, ρ2) are extracted in descending order of frequency (step S609).

  Next, the correction unit 225 has a second straight line that satisfies ρ1 = x · cos θ1 + y · sin θ1 and ρ2 = x · cos θ2 + y · sin θ2 in the xy coordinate system of the date stamp area (partial image). It is recognized that there are a first straight line portion located above the date in the date stamp and a second straight line portion located below the date on the straight line (step S610). In step S610, the correction unit 225 includes, for example, a first straight line portion located above the date on the first straight line, and a second straight line portion located below the date on the second straight line. Recognize that there is.

  Next, the correction unit 225 rotates the entire date stamp area cut out from the form image so that the first straight line portion and the second straight line portion are horizontal in the image (step S611). In the process of step S611, for example, the correction unit 225 calculates the average value θ = (θ1 + θ2) / 2 of the angles θ1 and θ2, and rotates the entire date stamp area by the angle −θ. When the process of step S611 is completed, the correction unit 225 ends the upright correction process.

  As described above, in the image processing apparatus 2 according to the present embodiment, after specifying the position of the date stamp in the image of the form, the inclination of the straight line portion of the date stamp with respect to the horizontal direction in the image is calculated, and the inside of the image is based on the inclination. Rotate the area marked with the date stamp. For this reason, in the image processing apparatus 2 of the present embodiment, even when the orientation of the date stamp impressed on the form is not upright, the orientation of the date stamp is automatically changed without the operator's operation. It is possible to perform a process of recognizing the date by correcting it in an upright direction.

FIG. 8 is a diagram illustrating an example of a form.
FIG. 8 shows an electricity bill transfer notification form which is an example of the form 8. On the upper side of the form 8, a character string 801 such as “Electricity transfer notification form” indicating the type of the form is printed. Below the character string 801, customer information 802 such as a customer number, application period, and name is printed. Below the customer information 802, transfer information 803 including information such as a customer number and a transfer amount is printed. On the lower left side of the transfer information 803, information 804 of the power company that issued the form 8 and a message 805 to the customer are printed. Further, a frame line 806 indicating a region for imprinting a date stamp indicating a date when the transfer procedure is accepted is printed on the lower right side of the transfer information 803.

  When the transfer procedure for the electricity bill is performed using the form 8 in FIG. 8, the clerk who has received the transfer stamps the date stamp indicating the date of the transfer in the area surrounded by the frame 806 of the form. . The image processing apparatus 2 according to the present embodiment acquires an image of the form 8 in which the date stamp is stamped in the area surrounded by the frame line 806, and recognizes the date included in the date stamp. However, the staff member performs the work of stamping the date stamp on the form 8 manually, and the staff member stamps the date stamp in a state where the stamp face cannot be seen. For this reason, the direction and position of the date stamp stamped on the form 8 are often different for each form.

  FIG. 9 is a diagram illustrating an example of a date stamp stamped on a form. FIG. 10 is a diagram showing another example of date stamps stamped on a form.

  FIG. 9A shows an example of the date stamp 9 stamped in an appropriate direction at an appropriate position in the area surrounded by the frame line 806 of the form 8. The date stamp 9 is generally divided into three regions by two linear portions 902 and 903 arranged in the vertical direction in the region surrounded by the contour portion 901, and the central region in the vertical direction of the three regions. The date 904 is displayed. Further, an item name 905 is displayed in the area above the date 904, and the name of the financial institution that accepted the transfer 906 is displayed in the area below the date 904.

  The form 8 is upright in the direction shown in FIG. For this reason, as shown in FIG. 9A, the date stamp 9 stamped on the form 8 includes a first straight line portion 902 located above the date and a second straight line portion 903 located below. The direction in which the stretching direction is parallel to the upper side (and the lower side) of the form 8 is upright. Thus, if the direction of the date stamp 9 stamped on the form 8 is upright, it is possible to correctly recognize the date 904 of the date stamp 9 from the image of the form 8. .

  FIG. 9B shows an example of the date stamp 9 in which the stamped position is appropriate but the orientation is not upright. When a staff member manually stamps the date stamp 9, the stamping position can be visually adjusted, but it is difficult to visually adjust the inclination of the stamp surface. For this reason, when the clerk manually stamps the date stamp 9, as shown in FIG. 9B, the extending direction of the straight portions 902 and 903 located above and below the date is (And the lower side) may be inclined. In this way, even when the direction of the date mark is slanted, if the angle formed by the upper side (and the lower side) of the form 8 and the extending direction of the linear portions 902 and 903 located above and below the date is small, The date 904 can be correctly recognized by character recognition. However, as shown in FIG. 9B, when the angle formed between the upper side (and the lower side) of the form 8 and the extending direction of the linear portions 902 and 903 located above and below the date is increased, It becomes difficult to recognize the date 904 correctly by recognition.

  FIG. 9C shows an example of the date mark 9 that is upright but protrudes from the area to be stamped. When the clerk stamps the date stamp 9 manually, a shift (variation) occurs in the stamping position depending on the clerk of the clerk who stamps the date stamp 9 or the situation (emergency or not) of the stamp. . For this reason, for example, as shown in FIG. 9C, a part of the date stamp 9 protrudes outside the area surrounded by the frame line 806, and within the area surrounded by the outline portion 901 of the date stamp 9. , A part of the frame line 806 printed on the form 8 may be included. In such a case, for example, when trying to specify the position of the date stamp 9 based on the heading or definition type of the form 8, the frame line 806 and the straight line portion 902 of the date stamp 9 are close to each other or overlap each other. Fail.

  FIG. 10 shows the date stamp 9 in which the area surrounded by the outline portion 911 is an octagon as another example of the date stamp 9. In the area surrounded by the outlines 901 and 911 in the date stamp 9, the date 904 is displayed at the center in the vertical direction, and there is information that can identify the clerk (financial institution) stamped above or below the date. It only has to be displayed. For this reason, there is a degree of freedom in the shape, outer shape, and the like of the region surrounded by the contour portion of the date stamp 9. That is, since the outer shape of the date stamp 9 is diverse, when the position of the date stamp 9 is specified from the image of the form 8 by pattern matching, it is necessary to prepare a large number of patterns (definitions).

  In view of such a situation, in the image processing apparatus 2 of the present embodiment, the features common to various date stamps having different outer shapes and colors, that is, the contour portions 901 and 911 of the date stamp 9 have an image horizontal direction and an image vertical direction. The position of the date stamp 9 in the image of the form 8 is specified by using a plurality of portions extending obliquely with respect to the direction. For this reason, the image processing apparatus 2 of the present embodiment performs the gradient extraction process (step S3) and the noise removal process (step S4) shown in the flowchart of FIG.

  FIG. 11 is a diagram for explaining a method of calculating the inclination. FIG. 12 is a diagram illustrating the calculation result of the inclination.

  FIG. 11 shows a form image 10 on which the date stamp 9 is stamped, an enlarged view of two partial areas Q1 and Q2 in the form image 10, and a filter 11. A partial area Q1 in the image 10 of the form is an area of 4 pixels × 5 pixels including a straight line (line segment) 1001 extending in the screen vertical direction of the form and the background. Further, the partial area Q2 in the image 10 of the form is an upwardly rising portion of the outline 901 of the date stamp 9 and extends obliquely with respect to each of the screen horizontal direction and the screen vertical direction. This is an area of 4 pixels × 5 pixels including a portion and a base.

  The numerical value described in each square in each of the enlarged partial areas Q1 and Q2 indicates the density of the corresponding pixel in the image 10 of the gray scaled form. For example, in the form image 10 converted to a gray scale of 256 gradations, the density of the pixels indicating the background is a large value close to 256, and the density of the pixels indicating the form elements (lines, characters, date stamps, etc.) is 0. A small value close to.

  When the gradient extraction process (step S3) is performed on the image 10 of the form, the image processing apparatus 2 has a gradient of pixels having a density equal to or higher than the threshold TH1 among the pixels in the image 10 as described above. Calculation is omitted (step S302; NO). That is, the leftmost five pixels and the rightmost five pixels in the partial region Q1, and the upper left white background pixel and the lower right white background pixel in the partial region Q2 are selected as processing targets, respectively. If so, the image processing apparatus 2 omits the calculation of the inclination.

When each of the black background pixels is selected as a processing target, the image processing apparatus 2 determines the inclination of the corresponding pixel based on the density of the adjacent eight pixels around the corresponding pixel and the filter 11. Calculate (step S303). At this time, as described above, the image processing apparatus 2 calculates the inclination of the corresponding pixel according to Equation (1). For example, the slope S _{u1, v1} of the pixel P _{u1, v1} (density 10) in the partial region Q1 in FIG. 11 is calculated by the following equation (2).

S _{u1, v1} = a1 · D _{u1-1, v1-1} + a2 · D _{u1, v1-1} + a3 · D _{u1 + 1, v1-1}
+ _A4・ D _{u1-1, v1} + a5 ・ D _{u1 + 1, v1}
+ _A6 · D _{u1-1, v1 + 1} + a7 · D _{u1, v1 + 1} + a8 · D _{u1 + 1, v1 + 1}
= A1 ・ 240 + a2 ・ 16 + a3 ・ 28
+ A4 ・ 250 + a5 ・ 15
+ A6 · 240 + a7 · 17 + a8 · 20 (2)

The coefficients a1 to a8 in the formula (2) are respectively given by the filter 11. In the filter 11 of FIG. 11, the central square among the 3 × 3 nine squares corresponds to the pixel currently being processed. That is, the filter 11 of FIG. 11, the pixel P _{u1 + 1, v1 + 1} at the lower right of the pixel P _u1, upper left pixel P _U1-1 of _v1, coefficients for _V1-1 a1, and the pixel P _{u1, v1} The coefficient a8 is set to “1”. Similarly, the filter 11 of FIG. 11, the pixel P _{u1, v1} of the upper right pixel P _U1-1, coefficients for _{v1 + 1} a3, and the pixel P _{u1, v1} lower left of the pixel P _U1-1, for _{v1 + 1} The coefficient a6 is set to “−1”. Furthermore, the filter 11 in FIG. 11 is a _{coefficient for} the pixels P _{u1, v1-1} , P _{u1, v + 1} , P _{u-1, v1} , and P _{u1, v1 + 1} of the upper, lower, left, and right pixels of the pixel P _{u1, v1.} This indicates that a2, a7, a4, and a5 are set to “0”. Therefore, when the calculated taking the value specified by the filter 11 to a1~a8 formula (2), inclination S _{u1, v1} pixel P _{u1, v1} (concentration 10) in the partial area Q1 in Fig. 11, As shown in FIG. 12, “−8” is obtained.

Further, for example, the slope S _{u2, v2} of the pixel P _{u2, v2} (density 20) in the partial region Q2 in FIG. 11 is “440” as shown in FIG.

  Further, although detailed description is omitted, the inclinations of the other black background pixels in the partial areas Q1 and Q2 are values as shown in FIG. It should be noted that “*” in the partial areas Q1 and Q2 in FIG.

  Further, the partial region Q2 in FIGS. 11 and 12 is a region including a pixel indicating a portion that rises to the right in the contour portion 901 of the date stamp 9. For this reason, the inclination calculated by applying the filter 11 of FIG. 11 to the pixels in the partial region Q2 is a positive value. On the other hand, although detailed explanation is omitted, when the slope is calculated by applying the filter 11 of FIG. 11 to the pixel indicating the downward-sloping part in the contour portion 901 of the date stamp 9, a negative value is obtained. Value. Therefore, in the determination process of step S304 performed after the process of calculating the inclination (step S303), it is determined whether or not the absolute value of the calculated inclination is equal to or greater than the threshold value TH2.

  As described above, when the inclination of the pixel indicating the form element (line, character, date stamp 9 or the like) in the form image 10 is calculated, the extending direction of the form elements is the horizontal direction or the vertical direction in the image 10. The inclination of the pixel indicating the portion substantially parallel to the pixel is clearly between the inclination of the pixel indicating the portion of the form element in which the extending direction between the horizontal direction and the vertical direction is greater than or equal to the threshold value. There is a big difference. For this reason, in the gradient extraction process (step S3) performed by the image processing apparatus 2 of the present embodiment, only pixels whose gradient value is greater than or equal to the threshold value TH2 are extracted from the pixels for which the gradient is calculated. The pixel information is held (step S305). Therefore, the gradient extraction result by the gradient extraction process is, for example, the extraction result 13 in FIG.

FIG. 13 is a diagram showing the result of the gradient extraction.
The extraction result 13 in FIG. 13 indicates in black the pixels that hold the inclination among the pixels in the form image 10. The four arc-shaped curves 1301, 1302, 1303, and 1304 located at the lower right corner in the extraction result 13 indicate the distribution of pixels in which the inclination in the contour portion 901 of the date stamp 9 is equal to or greater than the threshold value TH2. ing. In addition, two straight lines 1305 and 1306 that are located at the lower right corner in the extraction result 13 are a straight line portion 902 that is located above the date 904 of the date stamp 9 and a straight line portion 903 that is located below. The pixel distribution shown is shown (see FIG. 11).

  Further, the four arc-shaped curves 1307, 1308, 1309, and 1310 located at the lower side in the extraction result 13 are frame lines surrounding the message 804 (see FIG. 8) located at the lower side of the form 8, respectively. 2 shows the distribution of pixels indicating the corners in FIG. Further, the four arc-shaped curves 1311, 1312, 1313, and 1314 located in other parts of the extraction result 13 and the other arc-shaped curves are the other elements (frame line and character) in the form 8, respectively. ) In FIG. 6 shows a distribution of pixels having an inclination equal to or higher than a threshold value TH2. That is, when the gradient extraction process of step S3 is performed, the stretching direction in the form element different from the outline portion 901 and the straight line portions 902 and 903 of the date stamp 9 is the image as shown in the extraction result 13 of FIG. The inclination of the pixel indicating the portion that is oblique with respect to the horizontal direction and the image vertical direction is also extracted. For this reason, in the image processing performed by the image processing apparatus 2 of the present embodiment, as described above, the noise extraction processing (step S4) is performed after the gradient extraction processing, and the elements of the date stamp 9 (the contour portion 901 and the straight portion) are processed. Pixels indicating elements other than (902, 903, etc.) are excluded. In the noise removal process, as described above, first, a grouping process (step S401) is performed to group a plurality of pixels having the inclination in the extraction result 13 into groups for each successive pixel. Here, as a specific example of the grouping process, grouping for pixels in the region 1320 in the extraction result 13 of FIG. 13 will be described with reference to FIGS. 14A and 14B.

  FIG. 14A is a diagram (part 1) illustrating a method for grouping pixels. FIG. 14B is a diagram (part 2) illustrating a method for grouping pixels.

  For example, the image processing apparatus 2 performs processing according to the flowchart of FIG. 6 as the grouping processing. At this time, the image processing apparatus 2 performs raster scan in order from the pixel at the upper left corner in the image 10 of the form, and assigns a group number to the pixel holding the inclination.

14A shows the group number assignment status at the time when the pixel _{Pu1, v1} in the partial area 1320 ′ corresponding to the area 1320 in FIG. 13 in the form image 10 is selected as a processing target. Show. In FIG. 14A (a), the white background square is a pixel holding the slope, and the shaded square is a pixel not holding the slope. When selecting that the pixel of the processing by the raster scan, and the pixel located on the left side of the pixel P _{u1, v1,} the pixels included in the upper side of pixel columns than the pixel row including the pixel P _{u1, v1,} selected It is. Therefore, when the pixel P _{u1, v1} is selected, as shown in FIG. 14A (a), the inclination included in the pixel column above the pixel column including the pixel P _{u1, v1} is maintained. A group number is assigned to each pixel. Then, in allocating a group number to the pixel P _{u1, v1,} first, the upper left pixel P _{u1, v1,} on, whether the upper right, and the pixel group number allocated to either the left adjacent Is determined (step S401d). In the example shown in (a) of FIG. 14A, group numbers are assigned to the upper left adjacent pixel and the upper right adjacent pixel of the pixel _{Pu1, v1} . Further, the group number of the upper left adjacent pixel of the pixel _{Pu1, v1} is “1”, and the group number of the upper right adjacent pixel is “3”. Therefore, the image processing apparatus 2 assigns “1”, which is the smallest group number, to the pixels _{Pu1, v1} as shown in FIG. 14A (b) (step S401e).

After assigning group numbers to the pixels P _{u1, v1} , the image processing apparatus 2 has the same group number for the upper left adjacent pixel of the pixel P _{u1, v1} and the group number of the upper right adjacent pixel. It is determined whether or not there is (step S401f). In the example shown in (b) of FIG. 14A, the group numbers of the two pixels are different. Therefore, the image processing apparatus 2 registers “1” as the replacement number for the group number “3” in the replacement table 14 as shown in FIG. 14A (c) (step S401g).

  After that, although detailed description is omitted, when the loop processing in the grouping processing (steps S401b to S401j in FIG. 6) is finished, as shown in FIG. Is assigned a group number. After completing the loop processing (steps S401b to S401j in FIG. 6), the image processing apparatus 2 updates the group number allocated to the pixel based on the replacement table 14 (step S401k). Since the replacement table 14 includes information for replacing the group number “3” with “1”, the image processing apparatus 1 updates the group number of the pixel with the group number “3” to “1” ( (See (e) of FIG. 14B).

  When the grouping process (step S401) for grouping pixels having the inclination is finished, the image processing apparatus 2 performs the first loop process and the second loop process in the flowchart of FIG. Among the groups obtained by the conversion process, a group that does not satisfy the condition of the outline part of the date stamp is excluded from the group candidates indicating the outline part of the date stamp.

FIG. 15 is a diagram for explaining a noise removal method.
FIG. 15A shows a processing result 15 when the grouping process is performed based on the gradient extraction result 13. Four arc-shaped curves 1501, 1502, 1503, and 1504 located at the lower right corner in the processing result 15 indicate groups of pixels in which the inclination in the contour portion 901 of the date stamp 9 is equal to or greater than the threshold value TH2. ing. In addition, two straight lines 1505 and 1506 that are located at the lower right corner in the processing result 15 are a straight line portion 902 that is located above the date 904 of the date stamp 9 and a straight line portion 903 that is located below. The group of pixels shown is shown (see FIG. 11).

  In addition, the four arc-shaped curves 1507, 1508, 1509, and 1510 located at the lower side in the processing result 15 are frame lines surrounding the message 804 (see FIG. 8) located at the lower side of the form 8, respectively. The pixel group which shows the corner | angular part in is shown. Further, the four arc-shaped curves 1511, 1512, 1513, and 1514 located in other parts of the processing result 15 and the other arc-shaped curves are the other elements (frame line and character) in the form 8, respectively. ) Of the pixels having an inclination equal to or greater than the threshold value TH2. When the first loop process (steps S402 to S405 in FIG. 5) is performed on the processing result 15 as described above, a group whose vertical and horizontal lengths are shorter than the threshold value TH3 is selected from a group candidate indicating a date stamp outline. Excluded. The threshold value TH3 is a value shorter than the vertical and horizontal lengths in the group indicating the outline of the date stamp. Therefore, when the first loop processing is performed, for example, as shown in FIG. 15B, pixel groups 1511, 1512, 1513, and 1514 indicating the corners of the frame line in the form are displayed with date stamps. It is excluded from the group candidate showing the outline.

  When the first loop process ends, the image processing apparatus 2 performs a second loop process (steps S406 to S409 in FIG. 5). In the second loop processing, for each group, it is determined whether there are two or more valid groups around (step S407), and a group having only one valid group or a group having no valid group is defined as a date stamp outline. Are excluded from the group candidates indicating the part (step S408). In the determination processing in step S407, as described above, another effective group that partially or entirely is included in the search region that is set based on the distance between the plurality of groups that indicate the outline of the date stamp is searched. . Therefore, for example, there is only one set of the group 1510 that exists around the group 1507 corresponding to the form frame line in FIG. Similarly, each of the other effective groups existing around the groups 1508, 1509, and 1510 corresponding to the form frame line is one set. On the other hand, the groups 1501, 1502, 1503, and 1504 corresponding to the outline portion of the date stamp, and the groups 1505 and 1506 corresponding to the straight line portion, respectively, have two or more other valid groups existing around. Become. Therefore, by performing the second loop processing, the groups 1507, 1508, 1509, and 1510 corresponding to the form border are excluded from the group candidates indicating the date stamp outline. That is, when the noise removal processing is completed, only the groups 1501, 1502, 1503, and 1504 corresponding to the outline portion of the date mark in FIG. 15B and the groups 1505 and 1506 corresponding to the straight line portion are displayed. It remains as a candidate for a group that represents an outline. Therefore, in the image processing apparatus 2 of the present embodiment, the position of the date stamp automatically stamped on the form is automatically based on the position in the image of the group obtained by performing the gradient extraction process and the noise removal process. Can be specified.

  Furthermore, in the image processing apparatus 2 of the present embodiment, after specifying the position of the date stamp in the image 10 of the form, the inclination of the date stamp is corrected so that the direction of the date stamp is upright.

FIG. 16 is a diagram for explaining a method for calculating the inclination of the date stamp.
FIG. 16A shows an xy coordinate system of three pixels r1, r2, and r3 among a plurality of pixels indicating the first straight line portion 902 in the date stamp 9 cut out from the image 10 of the form. The coordinates of are shown. The x direction in the xy coordinate system is parallel to the u direction in the image, and the positive direction is the same as the u direction. On the other hand, the y direction in the xy coordinate system is parallel to the v direction in the image, and the positive direction is the direction opposite to the v direction.

  When calculating the inclination of the date stamp, as described above, the Hough transform is performed on the pixel indicating the element of the date stamp 9 in the region (partial image) cut out from the image 10 of the form. In the Hough transform, a distribution of ρ that satisfies the following formula (3) is calculated for each pixel indicating the element of the date stamp 9 within the range of 0 ≦ θ <2π.

    ρ = x · cosθ + y · sinθ (3)

  In the equation (3), x and y are the x coordinate and y coordinate in the xy coordinate system of the pixel selected as the processing target, respectively. For example, when the pixel r1 at the coordinates (x1, y1) in FIG. 16A is selected as the processing target, the image processing apparatus 2 satisfies the following formula (4) within the range of 0 ≦ θ <2π. The distribution of is calculated.

    ρ = x1 · cosθ + y1 · sinθ (4)

  When the relationship between θ and ρ with respect to the coordinates (x1, y1) of the pixel r1 calculated by the equation (4) is plotted in the θ-ρ coordinate system, for example, a curve as shown in FIG. Become. Although not described in detail, the relationship between θ and ρ with respect to the coordinates (x2, y2) of the pixel r2 and the relationship between θ and ρ with respect to the coordinates (x3, y3) of the pixel r3 are expressed as θ− When plotted in the ρ coordinate system, for example, the curves shown in FIG. Here, when the point (θ1, ρ1) that is the intersection of the three curves in the θ-ρ coordinate system is extracted and the following equation (5) is generated, the straight line that satisfies the equation (5) in the xy coordinate system is This is a straight line passing through the three pixels r1, r2, and r3.

    ρ1 = x · cos θ1 + y · sin θ1 (5)

  Also, Equation (5) can be rewritten as Equation (6).

    y = (− cos θ1 / sin θ1) · x + (ρ1 / sin θ1)

  That is, by deriving Equation (5) by the Hough transform, as shown in FIG. 16B, the inclination of the first linear portion 902 located above the date 904 in the date stamp 9 with respect to the x direction. The corner can be calculated.

  Further, the same procedure is used to derive the expression ρ2 = x · cos θ2 + y · sin θ2 indicating a straight line including each point on the second straight line portion 903 located below the date 904 in the date stamp 9 to obtain the second It is possible to calculate the inclination angle of the straight line portion with respect to the x direction. Since the first straight line portion 902 and the second straight line portion 903 in the date stamp 9 are substantially parallel, the values indicating the slopes in the two mathematical expressions calculated using the Hough transform are substantially the same.

  Thus, by calculating the inclination angles of the first straight line portion 902 and the second straight line portion 903 that extend in the horizontal direction when the date stamp 9 is upright, based on the Hough transform, the date The inclination angle θ of the mark 9 is known. Therefore, by rotating the date stamp cut out from the image 10 of the form by −θ, the inclination angle of the first straight line portion 902 and the second straight line portion 903 becomes 0 degrees, and the direction of the date mark becomes erect. .

FIG. 17 is a diagram for explaining how the date stamp is corrected upright.
FIG. 17A shows an example of a region Q3 when the region where the date stamp 9 is stamped is cut out from the image 10 of the form. For example, the region Q3 is a rectangular region having a size including all of the groups including the pixels indicating the outline portion of the date stamp obtained by the gradient extraction processing and the noise removal processing.

  After cutting out the region Q3 from the image 10 of the form, the image processing apparatus 2 uses the Hough transform to express the equation ρ1 = x · cos θ1 + y · sin θ1 indicating the straight line including the first straight line part 902 in the date stamp 9, and A mathematical expression ρ2 = x · cos θ2 + y · sin θ2 representing a straight line including two straight line portions 903 is derived. Thereafter, as shown in FIG. 17B, the image processing apparatus 2 calculates the inclination angle θ of the date stamp in the clipped region Q3 based on the derived two mathematical expressions.

  After calculating the inclination angle θ, the image processing apparatus 2 performs a conversion process of rotating the cut-out region Q3 by an angle −θ as shown in FIG. Thereby, the direction of the date stamp 9 becomes erect. Therefore, the date 904 of the date stamp 9 can be correctly recognized by performing character recognition on the area surrounded by the first straight line portion 902, the second straight line portion 903, and the contour portion 901 in the date stamp 9. It becomes.

  As described above, in the image processing apparatus 2 according to the present embodiment, the date stamp 9 in the form image 10 includes a plurality of portions in which the angle formed between the horizontal direction and the vertical direction of the image is equal to or greater than the threshold value. Is used to specify the position of the date stamp 9 in the image 10. Therefore, in the image processing apparatus 2 of the present embodiment, the position of the date stamp 9 is automatically specified regardless of the combination of the outline (outer shape), size, color, and the like of the date stamp 9 in the form image 10. It becomes possible. That is, according to the image processing apparatus 2 of the present embodiment, it is possible to quickly and flexibly cope with diversification of date stamps. Therefore, for example, the operator of the image processing apparatus 2 (image processing system 1) is freed from the task of defining diversified date stamp patterns and the task of selecting the area where the date stamp is imprinted in the image 10 of the form. The

  Furthermore, the image processing apparatus 2 according to the present embodiment calculates the inclination angle θ of the date stamp in the image 10 of the form, and sets the orientation of the date stamp so that the orientation of the date stamp is upright based on the inclination angle θ. to correct. For this reason, the image processing apparatus 2 can reduce erroneous recognition due to the inclination of the date stamp when the date included in the date stamp is recognized by character recognition. Therefore, for example, the operator of the image processing apparatus 2 (image processing system 1) is freed from the work of correcting the direction of the date stamp and the work of correcting the erroneously recognized date.

  Note that the flowchart of FIG. 3 is merely an example of image processing performed by the image processing apparatus 2 according to the present embodiment. The image processing performed by the image processing apparatus 2 of the present embodiment can be appropriately changed without departing from the gist described in the present embodiment. For example, the image processing performed by the image processing apparatus 2 of the present embodiment may include not only character recognition of the date stamp date but also character recognition of part or all of the character string in the form 8. Good. The image processing performed by the image processing apparatus 2 according to the present embodiment includes, for example, a process of generating and holding an image in which the corrected date stamp is combined with the form image 10 when the inclination of the date stamp is corrected. It may be included.

  Further, the flowchart of FIG. 4 is only an example of the gradient extraction process. The gradient extraction process is not limited to the process according to the flowchart of FIG. 4 and can be changed as appropriate without departing from the gist described in the present embodiment. For example, the gradient extraction process is performed by using a horizontal line and a vertical direction of a straight line in a form where the angle is greater than or equal to a predetermined angle (for example, 15 degrees or greater), and a tangential direction of the curve in the form is horizontal. A process of extracting a portion where the angle formed by each of the direction and the vertical direction is equal to or larger than a predetermined angle may be used.

  The flowcharts of FIGS. 5 and 6 are only examples of noise removal processing. The noise removal processing is not limited to the processing according to the flowcharts of FIGS. 5 and 6, and can be appropriately changed without departing from the gist described in the present embodiment. Furthermore, the flowchart of FIG. 7 is only an example of an upright correction process. The erecting correction process is not limited to the process according to the flowchart of FIG. 7, and can be appropriately changed without departing from the gist described in the present embodiment.

  In addition, the functional configuration of the image processing apparatus 2 according to the present embodiment is not limited to the configuration illustrated in FIG. For example, the character recognition dictionary 601 referred to by the recognition processing unit 230 may be stored in a storage unit (not shown) provided in the image processing apparatus 2. For example, when the form image 10 output by the scanner device 3 is a grayscale image or a binary image, the conversion unit 221 in the analysis unit 220 can be omitted. Further, for example, the gradient extraction unit 222 and the noise removal unit 223 in the analysis unit 220 are one processing unit that performs the gradient extraction process (step S3) and the noise removal process (step S4) as a series of processes. May be. Furthermore, the image processing apparatus 2 according to the present embodiment acquires a form image 10 provided via a portable recording medium such as a memory card or an optical disk or a network, and performs the above-described image processing. May be.

  In addition, the image processing apparatus 2 according to the present embodiment can be realized by a computer and a program executed by the computer. Hereinafter, the image processing apparatus 2 realized by a computer and a program will be described with reference to FIG.

FIG. 18 is a diagram illustrating a hardware configuration of a computer.
As illustrated in FIG. 18, the computer 20 includes a CPU (Central Processing Unit) 2001, a memory 2002, an auxiliary storage device 2003, an input device 2004, and a display device 2005. The computer 20 includes an input / output interface 2006, a communication control device 2007, and a medium driving device 2008. These elements 2001 to 2008 in the computer 20 are connected to each other by a bus 2010 so that data can be exchanged between the elements.

  The CPU 2001 controls the overall operation of the computer 20 by executing various programs including an operating system. For example, the CPU 2001 executes an image processing program including the processes of steps S1 to S8 illustrated in FIG. The gradient extraction process in the image processing program executed by the CPU 2001 is made into a subroutine as a process including the process of each step shown in the flowchart of FIG. Further, the noise extraction process in the image processing program executed by the CPU 2001 is made into a subroutine as a process including the process of each step shown in the flowcharts of FIGS. Further, the erecting correction process in the image processing program executed by the CPU 2001 is made into a subroutine as a process including the process of each step shown in the flowchart of FIG.

  The memory 2002 includes a ROM (Read Only Memory) and a RAM (Random Access Memory) not shown. In the ROM of the memory 2002, for example, a predetermined basic control program read by the CPU 2001 when the computer 20 is started is recorded in advance. The RAM of the memory 2002 is used as a working storage area as necessary when the CPU 2001 executes various programs. The RAM of the memory 2002 can be used for temporary storage of, for example, a form image, an inclination, a group number, and (θ, ρ) calculated by an upright correction process.

  The auxiliary storage device 2003 is, for example, a magnetic disk such as an HDD (Hard Disk Drive) and a nonvolatile memory such as a flash memory. The auxiliary storage device 2003 can store various programs executed by the CPU 2001, various data, and the like. The auxiliary storage device 2003 can be used for storing, for example, an image processing program including the processes of steps S1 to S8 shown in FIG. Further, the auxiliary storage device 2003 can be used for storing, for example, the form image 10 and the date recognition result of the date stamp.

  The input device 2004 is, for example, a keyboard device, a mouse device, a touch panel device, or the like. When the operator of the computer 20 performs a predetermined operation on the input device 2004, the input device 2004 transmits input information associated with the operation content to the CPU 2001. The input device 2004 corresponds to the input device 4 in the image processing system 1 of FIG.

  The display device 2005 is, for example, a liquid crystal display device. The display device 2005 can be used, for example, to display the operating state of the computer 20, the form image 10, the date recognition result of the date stamp, and the like. The display device 2005 corresponds to the display device 5 in the image processing system 1 of FIG.

  The input / output interface 2006 connects the computer 20 to electronic components and other electronic devices (for example, the scanner device 3). The input / output interface 2006 can also be used to connect a flash memory having a USB (Universal Serial Bus) standard connector, for example.

  The communication control device 2007 is a device that connects the computer 20 to a network and controls various communications between the computer 20 and other electronic devices via the network. The communication control device 2008 can be used, for example, to refer to a character recognition dictionary held by the server device 6 in the image processing system 1 of FIG.

  The medium driving device 2008 reads programs and data recorded in the portable storage medium 21 and writes data stored in the auxiliary storage device 2003 to the portable storage medium 21. As the portable storage medium 21, there is, for example, a Secure Digital (SD) standard memory card (flash memory). The portable recording medium 21 can be used for storing the image processing program, the form image 10, the date recognition result, and the like. When the computer 20 includes an optical disk drive that can be used as the medium drive device 2008, various optical disks that can be recognized by the optical disk drive can be used as the portable recording medium 21. Examples of the optical disc that can be used as the portable recording medium 21 include a Compact Disc (CD), a Digital Versatile Disc (DVD), and a Blu-ray Disc (Blu-ray is a registered trademark).

  When the operator operates the input device 2004 to input an image processing program start command, the computer 20 stores an image processing program including processing in steps S1 to S8 in FIG. 3 from the auxiliary storage device 2003 or the like. Read and execute. While executing the image processing program, the CPU 2001 functions (operates) as the analysis unit 220, the recognition processing unit 230, and the display control unit 240 in the image processing apparatus 2 illustrated in FIG. When acquiring the form image 10 from the scanner device 3, the CPU 2001 and the input / output interface 2006 function as the scanner control unit 210 in the image processing device 2 shown in FIG. Furthermore, when performing character recognition with reference to the character recognition dictionary 610 held by the server device 6 or the like different from the computer 20, the communication control device 2007 bears a part of the functions of the recognition processing unit 230.

  Note that the computer 20 operated as the image processing apparatus 2 does not have to include all of the elements 2001 to 2008 shown in FIG. 18, and some elements can be omitted depending on the application and conditions. For example, the computer 20 may be one in which the medium driving device 2008 and the communication control device 2007 are omitted.

DESCRIPTION OF SYMBOLS 1 Image processing system 2 Image processing apparatus 210 Scanner control part 220 Analysis part 221 Conversion part 222 Inclination extraction part 223 Noise removal part 224 Cutting-out part 225 Correction | amendment part 230 Recognition processing part 240 Display control part 250 Main control part 3 Scanner apparatus 4 Input device 5 Display device 6 Server device 601 Character recognition dictionary 7 Network 8 Form 9 Date stamp 10 Form image 11 Filter 13 Extraction result 14 Replacement table 15 Processing result 20 Computer 2001 CPU
2002 Memory 2003 Auxiliary storage device 2004 Input device 2005 Display device 2006 Input / output interface 2007 Communication control device 2008 Medium drive device 2010 Bus 21 Portable recording medium

Claims (6)

Out of the pixels indicating the element of the form in the form image, extract a pixel indicating a portion where the angle between the horizontal direction and the vertical direction of the form image in the element of the form is a predetermined angle or more,
Among the extracted pixels, a plurality of adjacent continuous pixels are grouped as one group,
From the group of pixels, extract a group in which the length and width in the image are equal to or greater than a predetermined threshold and there are two or more other groups whose distance is within a predetermined range,
An image processing program for causing a computer to execute a process of specifying an area including all of the extracted groups in an image of the form as an area having a date stamp. In the process of extracting the pixels, for each pixel indicating the element of the form, a density to be applied to each of a plurality of pixels located around the pixel and a coefficient to be applied to each of the densities of the plurality of pixels are determined. And calculating the sum of the product of the pixel density and the coefficient based on the filter
The image processing program according to claim 1, further comprising a process of extracting a pixel whose absolute value of the calculated sum is equal to or greater than a predetermined threshold. After identifying the area where the date stamp is stamped,
Cut out the area where the date stamp is stamped in the image of the form,
Calculate the inclination angle of the straight line portion in the date stamp included in the region with respect to the horizontal direction in the cut out region,
3. The computer according to claim 1, further comprising: causing the computer to execute a process of rotating the cut-out area so that the extending direction of the linear portion in the date stamp is a horizontal direction based on the calculated inclination angle. Image processing program. After identifying the area where the date stamp is stamped,
The image processing program according to claim 1, further comprising: causing a computer to execute processing for recognizing a date included in the date stamp in an area where the date stamp is stamped by character recognition. Among the pixels indicating the element of the form in the form image, the computer indicates a pixel indicating a portion where the angle between the horizontal direction and the vertical direction of the form image in the form element is equal to or larger than a predetermined angle. Extract and
Among the extracted pixels, a plurality of adjacent continuous pixels are grouped as one group,
From the group of pixels, extract a group in which the length and width in the image are equal to or greater than a predetermined threshold and there are two or more other groups whose distance is within a predetermined range,
An image processing method comprising: executing a process of specifying an area including all of the extracted groups in the form image as an area having a date stamp. Out of the pixels indicating the element of the form in the form image, extract a pixel indicating a portion where the angle between the horizontal direction and the vertical direction of the form image in the element of the form is a predetermined angle or more, Among the extracted pixels, a plurality of adjacent continuous pixels are grouped as one group, and then the vertical and horizontal lengths in the image are equal to or greater than a predetermined threshold in the group of pixels, and the distance An analysis unit that extracts a group in which two or more other groups within a predetermined range exist, and identifies a region including all of the extracted groups in the image of the form as a region stamped with a date stamp An image processing apparatus comprising: