JP2009020867A - Image processing method, image processing apparatus, image forming apparatus, computer program, and recording medium - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an image processing method capable of stably performing an image similarity determination without degeading determination accuracy by a processing apparatus reduced in cost and scale, and to provide an image processing apparatus, an image forming apparatus, a computer program, and a recording medium. <P>SOLUTION: A cnt value is read as an address so as to determine whether an lbnum[cnt], which is the number of counted pixels of an object label, is lower than a first threshold value (S201). When the lbnum[cnt] is determined to be lower than the first threshold (YES in S201), data including the ibnum[cnt] is abandoned (S202). Then, the first threshold is changed. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to an image processing method for determining similarity between images, an image processing apparatus for executing the method, an image forming apparatus, a computer program for causing the image processing apparatus to execute the method, and a recording medium on which the computer program is recorded. In particular, the present invention relates to an image processing method, an image processing apparatus, an image forming apparatus, a computer program, and a recording medium that can improve the accuracy of similarity determination of an image with few feature points.

  Conventionally, input image data obtained by reading a document image with a scanner is compared with a registered image registered in advance to determine the degree of similarity between the two, and processing on the input image data (for example, copying) , Transmission, editing, etc.) are used.

  As a method for determining the similarity, for example, a keyword is extracted from an image using an OCR (Optical Character Reader) or the like and matching is performed based on the keyword, or the target image is limited to a form image having a ruled line, and matching is performed based on the characteristic of the ruled line. The method of performing is known (for example, patent document 1).

In Patent Document 2, a plurality of feature points are extracted from a digital image, a set of local feature points is determined for each extracted feature point, and a subset of feature points is selected from each determined set Then, as an amount to characterize each selected subset, invariants for geometric transformation are obtained based on a plurality of combinations of feature points in the subset, and feature amounts are calculated by combining the obtained invariants. A technique for searching for a document / image corresponding to the digital image by voting to a document / image in a database based on the calculated feature amount is disclosed (for example, Patent Document 2).
JP-A-8-255236 International Publication No. 2006/092957 Pamphlet

  FIG. 24 is an explanatory diagram showing the center of gravity of the connected component, and FIG. 25 is an explanatory diagram showing an example using the center of gravity as a feature point. The centroid of the connected component is not easily affected by noise, and its position does not change even when the target image rotates or translates. It is said that the image processing method for performing image similarity can realize a high-accuracy image similarity determination (see FIG. 24).

  However, in this method, since the number of centroids calculated based on the input document is reduced, the number of feature points may be reduced. In addition, if the feature amount is calculated using a small number of feature points, the feature points necessary for the calculation cannot be secured sufficiently, and there is a problem that the accuracy of the calculated feature amount itself is lowered.

  The present invention has been made in view of such circumstances, and an object of the present invention is to specify a plurality of pixel regions adjacent to each other that are determined to have the same pixel value, and to determine the number of pixels in the specified pixel region. Is equal to or greater than the first threshold value, when performing similarity determination between images using the feature amount calculated based on the feature points extracted from the specified pixel region, the score of the extracted feature points is the second threshold value. An image processing method, an image processing apparatus, an image forming apparatus, and a computer capable of stably processing image similarity determination without reducing the determination accuracy by changing the first threshold when To provide a program and a recording medium.

  Another object of the present invention is to specify a plurality of adjacent pixel regions that are determined to have the same pixel value, and whether or not the number of pixels in the specified pixel region is equal to or greater than a first threshold value. When determining a plurality of different first threshold values, determining whether or not each of the first threshold values is equal to or greater than the first threshold value, and determining that the threshold value is equal to or greater than the first threshold value, In addition to calculating points, the number of feature points is counted for each of a plurality of different first thresholds, and the images are calculated using feature amounts calculated based on the feature points extracted from the pixel area where the number of counted feature points is the largest. An image processing method, an image processing apparatus, an image forming apparatus, a computer program, and a recording medium capable of stably processing image similarity determination without reducing determination accuracy by performing similarity determination of It is in.

  Another object of the present invention is to divide a document image and to set a lower limit for the number of pixels in the connected region, that is, a first threshold value for each of the divided regions. Not only is it possible to set a threshold value so that the noise in the upper, lower, left and right edge areas of the image is not calculated as the center of gravity, but also in the important areas (for example, the central area) of the original image, An object of the present invention is to provide an image processing method, an image processing apparatus, an image forming apparatus, a computer program, and a recording medium that can set a threshold value for calculating the center of gravity and can secure a sufficient number of high-precision center of gravity.

  Another object of the present invention is to generate a region for selecting feature points, perform similarity determination between images using feature amounts calculated based on feature points located in the generated region, and extract them. Image processing capable of stably performing image similarity determination without reducing determination accuracy by changing the region for selecting the feature point when the number of feature points is equal to or less than a second threshold value A method, an image processing apparatus, an image forming apparatus, a computer program, and a recording medium are provided.

  Another object of the present invention is to read out feature points for each processing of feature points of interest by counting the number of feature points within a predetermined range (pixel block line) when extracting feature points. Since the mask size or the number of reference blocks can be changed, there is no need to repeatedly extract neighboring feature points, and the image processing method, the image processing apparatus, and the like, which can facilitate and speed up the processing, To provide an image forming apparatus, a computer program, and a recording medium.

  Another object of the present invention is to specify a plurality of adjacent pixel regions that are determined to have the same pixel value, and specify when the number of pixels in the specified pixel region is equal to or greater than a first threshold value. When performing similarity determination between images using a feature amount calculated based on a feature point extracted from a pixel area extracted from the extracted pixel area, the number of extracted feature points is a second threshold value. By changing the first threshold value or the first threshold value and the range of the set area when the following is true, it is possible to stably perform the image similarity determination without reducing the determination accuracy. An image processing method, an image processing apparatus, an image forming apparatus, a computer program, and a recording medium are provided.

  Another object of the present invention is to specify a plurality of adjacent pixel regions that are determined to have the same pixel value, and whether or not the number of pixels in the specified pixel region is equal to or greater than a first threshold value. When determining a plurality of different first threshold values, determining whether or not each of the first threshold values is equal to or greater than the first threshold value, and determining that the threshold value is equal to or greater than the first threshold value, While calculating points, the number of feature points is counted for each of a plurality of different first threshold values, and the feature points extracted from the pixel region having the largest number of feature points are within the set region range. An image processing method capable of stably processing image similarity determination without reducing determination accuracy by performing similarity determination between images with feature amounts calculated based on the feature points of the largest number of positions, Image processing apparatus, image forming apparatus, computer And to provide a program and a recording medium.

  Another object of the present invention is to extract a feature point by changing a size or the number of blocks of the pixel block by defining a region from which the feature point is extracted by a pixel block composed of one or a plurality of pixels. An object of the present invention is to provide an image processing method, an image processing apparatus, and an image forming apparatus that can change a region to be processed.

  In order to solve the above-described problem, an image processing method according to the present invention specifies a plurality of pixel regions adjacent to pixels that are determined to have the same pixel value from a binary image, and coordinates of the pixels of the specified pixel region. In an image processing method for extracting feature points of a pixel area based on a value, calculating a feature amount indicating an image feature based on the extracted feature point, and determining similarity between images based on the calculated feature amount , Counting the number of pixels in the specified pixel region, determining whether the number of pixels in the specified pixel region is equal to or greater than a first threshold, and determining the number of pixels in the specified pixel region to be the first threshold When it is determined as above, the feature points of the pixel region are calculated, the number of feature points is counted, it is determined whether or not the counted number of feature points is equal to or less than a second threshold, and the counted features When it is determined that the number of points exceeds the second threshold The feature amount is calculated based on the feature points extracted from the pixel region, and the first threshold value is changed when it is determined that the number of counted feature points is equal to or less than a second threshold value. .

  In addition, the image processing method according to the present invention specifies a plurality of pixel regions adjacent to pixels that are determined to have the same pixel value from a binary image, and based on the coordinate values of the pixels in the specified pixel region. In an image processing method for extracting feature points of a pixel area, calculating a feature amount indicating an image feature based on the extracted feature point, and determining similarity between images based on the calculated feature amount, the specified pixel region When determining whether or not the number of pixels in the specified pixel region is greater than or equal to the first threshold, a plurality of different first thresholds are set, each of which is greater than or equal to the first threshold. If it is determined whether or not it is greater than or equal to the first threshold, the feature points of the pixel area are calculated, and the number of feature points is counted for each of a plurality of different first thresholds. Extracted from the pixel region that has the largest number of feature points. And calculating a feature quantity based on the feature points.

  In addition, the image processing method according to the present invention specifies a plurality of pixel regions adjacent to pixels that are determined to have the same pixel value from a binary image, and based on the coordinate values of the pixels in the specified pixel region. In an image processing method for extracting feature points of a pixel area, calculating a feature amount indicating an image feature based on the extracted feature point, and determining similarity between images based on the calculated feature amount, a region of an original image The first threshold value is set for each divided area, the number of pixels in the specified pixel area is counted in each divided area, and specified in the divided area. It is determined whether the number of pixels in the pixel area is equal to or greater than a first threshold in the divided area, and the number of pixels in the pixel area specified in the divided area is the first in the divided area. Above threshold If it is determined, the feature points of the pixel region are calculated, the number of feature points is counted, it is determined whether or not the counted number of feature points is equal to or less than a second threshold, and the number of feature points counted Is calculated based on the feature points extracted from the pixel region, and the number of counted feature points is determined to be less than or equal to the second threshold The first threshold value is changed.

  In addition, the image processing method according to the present invention specifies a plurality of pixel regions adjacent to pixels that are determined to have the same pixel value from a binary image, and based on the coordinate values of the pixels in the specified pixel region. Extracted feature points in an image processing method that extracts feature points of a pixel area, calculates feature amounts indicating image features based on the extracted feature points, and performs similarity determination between images based on the calculated feature amounts The feature point of the neighborhood located in the peripheral region of is determined, the number of feature points of the decided neighborhood is counted, it is determined whether or not the score of the feature point is less than or equal to the second threshold value, and counted When it is determined that the number of feature points exceeds the second threshold, the feature amount is calculated based on the neighboring feature points, and when it is determined that the counted number is equal to or less than the second threshold, the peripheral region The range is changed.

  In addition, the image processing method according to the present invention specifies a plurality of pixel regions adjacent to pixels that are determined to have the same pixel value from a binary image, and based on the coordinate values of the pixels in the specified pixel region. Extract feature points in an image processing method that extracts feature points of a pixel area, calculates feature amounts indicating image features based on the extracted feature points, and determines similarity between images based on the calculated feature amounts In this case, the number of feature points within a predetermined range is counted, and feature points included in the region to be processed are extracted from the counted feature points, and whether the score of the extracted feature points is smaller than the third threshold value If the number of extracted feature points is determined to be greater than or equal to the third threshold value, the feature amount is calculated based on neighboring feature points located in the peripheral region of the extracted feature points. , It is determined that the extracted score is smaller than the third threshold value. If and changes the range of the peripheral region.

  In addition, the image processing method according to the present invention specifies a plurality of pixel regions adjacent to pixels that are determined to have the same pixel value from a binary image, and based on the coordinate values of the pixels in the specified pixel region. In an image processing method for extracting feature points of a pixel area, calculating a feature amount indicating an image feature based on the extracted feature point, and determining similarity between images based on the calculated feature amount, the specified pixel region The number of pixels in the specified pixel region is counted to determine whether or not the number of pixels in the specified pixel region is equal to or greater than the first threshold, and the number of pixels in the specified pixel region is determined to be equal to or greater than the first threshold. In this case, the feature points of the pixel region are extracted, neighboring feature points located in the surrounding region of the extracted feature points are determined, the number of feature points in the determined neighborhood is counted, and the score of the counted feature points is Determine whether it is below the second threshold and count When it is determined that the number of feature points exceeds the second threshold, the feature amount is calculated based on the neighboring feature points, and it is determined that the counted number of feature points is equal to or less than the second threshold Changes the first threshold value, or the first threshold value and the range of the peripheral region.

  In addition, the image processing method according to the present invention specifies a plurality of pixel regions adjacent to pixels that are determined to have the same pixel value from a binary image, and based on the coordinate values of the pixels in the specified pixel region. In an image processing method for extracting feature points of a pixel area, calculating a feature amount indicating an image feature based on the extracted feature point, and determining similarity between images based on the calculated feature amount, the specified pixel region When determining whether or not the number of pixels in the specified pixel region is greater than or equal to the first threshold, a plurality of different first thresholds are set, each of which is greater than or equal to the first threshold. If it is determined whether or not it is greater than or equal to the first threshold, the feature points of the pixel area are calculated, and the number of feature points is counted for each of a plurality of different first thresholds. The feature points of the pixel area that has the largest number of feature points The feature points in the vicinity of the extracted feature points are determined, and the number of feature points in the determined neighborhood is counted, and whether or not the scored feature points are equal to or less than the second threshold If the number of feature points counted is determined to exceed a second threshold, the feature amount is calculated based on the neighboring feature points, and the number of counted feature points is less than or equal to a second threshold. If it is determined that there is, the range of the surrounding area is changed.

  Further, in the image processing method according to the present invention, the peripheral area is configured by a pixel block including one or a plurality of pixels of a binary image, and the change of the range of the peripheral area is the size or block of the pixel block. It is characterized by being performed by changing the number.

  In addition, the image processing apparatus according to the present invention specifies a plurality of pixel regions adjacent to pixels that are determined to have the same pixel value from a binary image, and based on the coordinate values of the pixels in the specified pixel region, In the image processing apparatus that extracts feature points of the pixel area, calculates a feature amount indicating the feature of the image based on the extracted feature point, and performs similarity determination between the images based on the calculated feature amount, the specified pixel region Means for counting the number of pixels, means for determining whether the number of pixels in the specified pixel area is equal to or greater than a first threshold, and the number of pixels in the identified pixel area is greater than or equal to the first threshold A means for calculating the feature points of the pixel region and counting the number of feature points, and a means for determining whether or not the counted number of feature points is equal to or less than a second threshold. And the number of feature points counted has a second threshold value. If it is determined that the number of feature points is less than or equal to a second threshold value, the first threshold value is changed. It is made to do so.

  In addition, the image processing apparatus according to the present invention specifies a plurality of pixel regions adjacent to pixels that are determined to have the same pixel value from a binary image, and based on the coordinate values of the pixels in the specified pixel region, In the image processing apparatus that extracts feature points of the pixel area, calculates a feature amount indicating the feature of the image based on the extracted feature point, and performs similarity determination between the images based on the calculated feature amount, the specified pixel region When determining whether or not the number of pixels is equal to or more than the first threshold value, a plurality of different first threshold values are set, each of which is greater than or equal to the first threshold value. A determination means for determining whether or not there is, and when it is determined that the value is equal to or greater than the first threshold value, the feature points of the pixel region are calculated and the number of feature points is counted for each of a plurality of different first threshold values The number of feature points counted is Characterized in that are to calculate the characteristic amount based on the feature points extracted from the pixel region to be a multi.

  In addition, the image processing apparatus according to the present invention specifies a plurality of pixel regions adjacent to pixels that are determined to have the same pixel value from a binary image, and based on the coordinate values of the pixels in the specified pixel region, In an image processing apparatus that extracts feature points of a pixel area, calculates a feature amount indicating an image feature based on the extracted feature point, and performs similarity determination between images based on the calculated feature amount, a region of an original image Divided into a plurality of areas, means for setting a first threshold for each divided area, means for counting the number of pixels in the specified pixel area in each divided area, and Means for determining whether the number of pixels in the pixel area specified in the divided area is equal to or greater than a first threshold value in the divided area; and the number of pixels in the pixel area specified in the divided area Was Means for calculating the feature points of the pixel region and counting the number of feature points, and whether the counted number of feature points is less than or equal to the second threshold value Means for determining whether or not the number of feature points counted exceeds a second threshold value, the feature amount is calculated based on the feature points extracted from the pixel region, and When it is determined that the number is equal to or smaller than the second threshold value, the first threshold value is changed.

  In addition, the image processing apparatus according to the present invention specifies a plurality of pixel regions adjacent to pixels that are determined to have the same pixel value from a binary image, and based on the coordinate values of the pixels in the specified pixel region, Extracted feature points in an image processing apparatus that extracts feature points of a pixel area, calculates feature amounts indicating image features based on the extracted feature points, and performs similarity determination between images based on the calculated feature amounts Means for determining neighboring feature points located in the peripheral region of the image, means for counting the number of feature points in the vicinity, and determining whether the score of the counted feature points is equal to or less than a second threshold value And calculating the feature amount based on the neighboring feature points, and the score of the counted feature points is the second threshold value when the number of feature points counted is determined to exceed the second threshold value. If it is determined that: Characterized in that a are to be changed.

  In addition, the image processing apparatus according to the present invention specifies a plurality of pixel regions adjacent to pixels that are determined to have the same pixel value from a binary image, and based on the coordinate values of the pixels in the specified pixel region, Extracting feature points in a pixel area, calculating feature amounts indicating image features based on the extracted feature points, and extracting feature points in an image processing apparatus that determines similarity between images based on the calculated feature amounts In this case, the means for counting the number of feature points within a predetermined range, the means for extracting the feature points included in the region to be processed from the counted feature points, and the score of the extracted feature points from the third threshold Means for determining whether or not the number of extracted feature points is greater than or equal to a third threshold value, based on neighboring feature points located in the peripheral region of the extracted feature points The feature amount is calculated, and the number of points extracted is Characterized in that when it is determined that less than the threshold value of are so as to change the range of the peripheral region.

  In addition, the image processing apparatus according to the present invention specifies a plurality of pixel regions adjacent to pixels that are determined to have the same pixel value from a binary image, and based on the coordinate values of the pixels in the specified pixel region, In the image processing apparatus that extracts feature points of the pixel area, calculates a feature amount indicating the feature of the image based on the extracted feature point, and performs similarity determination between the images based on the calculated feature amount, the specified pixel region Means for counting the number of pixels, means for determining whether the number of pixels in the specified pixel area is equal to or greater than a first threshold, and the number of pixels in the identified pixel area is greater than or equal to the first threshold. When it is determined that there is a means for extracting feature points of the pixel area, a means for determining neighboring feature points located in a peripheral area of the extracted feature points, and counting the number of feature points in the determined neighborhood Means and the number of feature points counted Means for determining whether or not the number of feature points counted is greater than the second threshold, the feature amount is calculated based on the neighboring feature points, The first threshold value, or the first threshold value and the range of the surrounding area are changed when it is determined that the number of feature points counted is equal to or less than the second threshold value. .

  In addition, the image processing apparatus according to the present invention specifies a plurality of pixel regions adjacent to pixels that are determined to have the same pixel value from a binary image, and based on the coordinate values of the pixels in the specified pixel region, In the image processing apparatus that extracts feature points of the pixel area, calculates a feature amount indicating the feature of the image based on the extracted feature point, and performs similarity determination between the images based on the calculated feature amount, the specified pixel region The means for counting the number of pixels and a plurality of different first threshold values are set when determining whether or not the number of pixels in the specified pixel region is equal to or greater than the first threshold value. When it is determined that it is equal to or greater than the first threshold value, the feature point of the pixel region is calculated and the number of feature points is set to a plurality of different first threshold values. The counting means and the number of feature points counted are Means for extracting the feature points of the pixel area to be, means for determining neighboring feature points located in the peripheral area of the extracted feature points, means for counting the number of feature points in the determined neighborhood, and counting Means for determining whether or not the number of feature points is equal to or less than a second threshold, and when it is determined that the counted number of feature points exceeds the second threshold, based on the nearby feature points The feature amount is calculated, and when it is determined that the number of counted feature points is equal to or smaller than the second threshold value, the range of the peripheral region is changed.

  In the image processing apparatus according to the present invention, the peripheral area is configured by a pixel block including one or a plurality of pixels of a binary image, and the change of the range of the peripheral area is the size or block of the pixel block. It is characterized by being performed by changing the number.

  An image forming apparatus according to the present invention includes the image processing apparatus according to the present invention and an image forming unit that forms an image processed by the image processing apparatus.

  In addition, the computer program according to the present invention causes a computer to specify a plurality of adjacent pixel regions that are determined to have the same pixel value from a binary image, and based on the coordinate values of the pixels in the specified pixel region. In a computer program for extracting feature points of the pixel area, calculating a feature amount indicating an image feature based on the extracted feature point, and determining similarity between images based on the calculated feature amount. A step of counting the number of pixels in the specified pixel region; a step of causing the computer to determine whether the number of pixels in the specified pixel region is equal to or greater than a first threshold; and When it is determined that the number of pixels is equal to or greater than the first threshold, the feature points of the pixel region are calculated and the number of feature points is counted. And causing the computer to determine whether or not the counted number of feature points is less than or equal to a second threshold, and if the computer determines that the number of counted feature points exceeds the second threshold Calculating the feature amount based on the feature points extracted from the pixel area, and changing the first threshold value when the computer determines that the number of feature points counted is equal to or less than a second threshold value. And executing the step.

  In addition, the computer program according to the present invention causes a computer to specify a plurality of adjacent pixel regions that are determined to have the same pixel value from a binary image, and based on the coordinate values of the pixels in the specified pixel region. In a computer program for extracting feature points of the pixel area, calculating a feature amount indicating an image feature based on the extracted feature point, and determining similarity between images based on the calculated feature amount. A step of counting the number of pixels in the specified pixel region, and when the computer determines whether or not the number of pixels in the specified pixel region is equal to or greater than the first threshold, a plurality of different first threshold values are set. A step of setting and determining whether or not each is greater than or equal to a first threshold, and if the computer determines that the threshold is greater than or equal to the first threshold, Calculating a feature point and counting the number of feature points for each of a plurality of different first threshold values, and causing the computer to perform a feature based on the feature point extracted from the pixel region having the largest number of counted feature points And a step of calculating a quantity.

  In addition, the computer program according to the present invention causes a computer to specify a plurality of adjacent pixel regions that are determined to have the same pixel value from a binary image, and based on the coordinate values of the pixels in the specified pixel region. In a computer program for extracting feature points of the pixel area, calculating a feature amount indicating an image feature based on the extracted feature point, and determining similarity between images based on the calculated feature amount. A step of dividing the region of the original image into a plurality of steps, a step of causing the computer to set a first threshold value for each of the divided regions, and a pixel region identified by the computer in each of the divided regions A step of counting the number of pixels in the divided area, and causing the computer to calculate the number of pixels in the divided area. And determining whether the number of pixels in the pixel area specified in the divided area is equal to or greater than the set first threshold value. Calculating feature points and counting the number of feature points, causing the computer to determine whether or not the counted number of feature points is equal to or less than a second threshold, and causing the computer to count the feature points The number of feature points is calculated based on the feature points extracted from the pixel area when the number of feature points exceeds the second threshold, and the computer counts the number of feature points below the second threshold. If it is determined that there is, the step of changing the first threshold value is executed.

  In addition, the computer program according to the present invention causes a computer to specify a plurality of adjacent pixel regions that are determined to have the same pixel value from a binary image, and based on the coordinate values of the pixels in the specified pixel region. In a computer program for extracting feature points of the pixel area, calculating a feature amount indicating an image feature based on the extracted feature point, and determining similarity between images based on the calculated feature amount. A step of determining neighboring feature points located in the peripheral region of the extracted feature points, a step of causing a computer to count the number of feature points in the determined neighborhood, and a step of causing the computer to determine the number of feature points counted. A step for determining whether or not the threshold value is equal to or less than a threshold value of 2, and a case where the computer determines that the number of feature points counted exceeds a second threshold value Calculating the feature amount based on the nearby feature points, and causing the computer to change the range of the surrounding area when it is determined that the number of counted feature points is equal to or less than a second threshold value. It is made to perform.

  In addition, the computer program according to the present invention causes a computer to specify a plurality of adjacent pixel regions that are determined to have the same pixel value from a binary image, and based on the coordinate values of the pixels in the specified pixel region. In a computer program for extracting feature points of the pixel area, calculating a feature amount indicating an image feature based on the extracted feature point, and determining similarity between images based on the calculated feature amount. When extracting feature points, the step of counting the number of feature points within a predetermined range, the step of causing the computer to determine whether the number of feature points extracted is smaller than a third threshold, and the computer extracting If it is determined that the number of feature points obtained is greater than or equal to the third threshold value, the feature points are determined based on the feature points in the vicinity located in the peripheral region of the extracted feature points. A step of calculating the amount, the computer, the extracted number is when it is determined to be smaller than the third threshold value, characterized in that and a step of changing the range of the peripheral region.

  In addition, the computer program according to the present invention causes a computer to specify a plurality of adjacent pixel regions that are determined to have the same pixel value from a binary image, and based on the coordinate values of the pixels in the specified pixel region. In a computer program for extracting feature points of the pixel area, calculating a feature amount indicating an image feature based on the extracted feature point, and determining similarity between images based on the calculated feature amount. A step of counting the number of pixels in the specified pixel region; a step of causing the computer to determine whether the number of pixels in the specified pixel region is equal to or greater than a first threshold; and If the number of pixels is determined to be greater than or equal to the first threshold, a feature point of the pixel region is extracted, and the computer extracts A step of determining neighboring feature points located in the peripheral area of the feature point; a step of causing the computer to count the number of points of the determined neighboring feature point; A step of determining whether or not the number of feature points counted is greater than the second threshold, and causing the computer to calculate the feature amount based on the nearby feature points when determined that the number of feature points exceeds the second threshold; And causing the computer to execute the step of changing the first threshold value or the first threshold value and the range of the surrounding area when it is determined that the number of feature points counted is equal to or less than the second threshold value. It is characterized by that.

  In addition, the computer program according to the present invention causes a computer to specify a plurality of adjacent pixel regions that are determined to have the same pixel value from a binary image, and based on the coordinate values of the pixels in the specified pixel region. In a computer program for extracting feature points of the pixel area, calculating a feature amount indicating an image feature based on the extracted feature point, and determining similarity between images based on the calculated feature amount. A step of counting the number of pixels in the specified pixel region, and when the computer determines whether or not the number of pixels in the specified pixel region is equal to or greater than the first threshold, a plurality of different first threshold values are set. A step of setting and determining whether or not each is greater than or equal to a first threshold, and if the computer determines that the threshold is greater than or equal to the first threshold, Calculating feature points, counting the number of feature points for each of a plurality of different first thresholds, and causing the computer to extract feature points of a pixel region having the largest number of feature points counted; A step of causing a computer to determine neighboring feature points located within the peripheral region of the extracted feature points; a step of causing the computer to score the number of feature points in the neighborhood; and A step of determining whether or not the threshold value is equal to or less than a threshold value of 2, and causing the computer to calculate the feature amount based on the neighboring feature points when it is determined that the number of counted feature points exceeds the second threshold value And a step of causing the computer to change the range of the peripheral area when it is determined that the counted number of feature points is equal to or less than the second threshold value. It is characterized in.

  Further, the recording medium according to the present invention causes a computer to specify a plurality of pixel areas adjacent to pixels determined to have the same pixel value from a binary image, and based on the coordinate values of the pixels in the specified pixel area. A computer program for extracting feature points of the pixel region, calculating a feature amount indicating the feature of the image based on the extracted feature point, and determining similarity between the images based on the calculated feature amount; In a recording medium readable by a computer, a step of causing the computer to count the number of pixels in the specified pixel region, and whether the number of pixels in the specified pixel region is equal to or greater than a first threshold value And when the computer determines that the number of pixels in the specified pixel area is equal to or greater than the first threshold, the feature point of the pixel area is determined. And counting the number of feature points, causing the computer to determine whether the counted number of feature points is less than or equal to a second threshold, and causing the computer to determine the number of feature points If it is determined that the threshold value of 2 is exceeded, the step of calculating the feature amount based on the feature points extracted from the pixel region and the computer determines that the number of feature points counted is equal to or less than the second threshold value. In this case, a computer program for executing the step of changing the first threshold value is recorded.

  Further, the recording medium according to the present invention causes a computer to specify a plurality of pixel areas adjacent to pixels determined to have the same pixel value from a binary image, and based on the coordinate values of the pixels in the specified pixel area. A computer program for extracting feature points of the pixel region, calculating a feature amount indicating the feature of the image based on the extracted feature point, and determining similarity between the images based on the calculated feature amount; In a recording medium readable by a computer, a step of causing the computer to count the number of pixels in the specified pixel region, and whether the number of pixels in the specified pixel region is equal to or greater than a first threshold value When determining whether to set a plurality of different first threshold values, and determining whether or not each of the first threshold values is equal to or greater than the first threshold value, If it is determined that the number of feature points of the pixel region is calculated, a step of counting the number of feature points for each of a plurality of different first threshold values, and a computer having the largest number of feature points counted. The computer program which performs the step which calculates a feature-value based on the feature point extracted from the said pixel area is recorded.

  Further, the recording medium according to the present invention causes a computer to specify a plurality of pixel areas adjacent to pixels determined to have the same pixel value from a binary image, and based on the coordinate values of the pixels in the specified pixel area. A computer program for extracting feature points of the pixel region, calculating a feature amount indicating the feature of the image based on the extracted feature point, and determining similarity between the images based on the calculated feature amount; In a recording medium readable by a computer, the step of causing the computer to divide the region of the original image into a plurality of steps, the step of causing the computer to set a first threshold value for each of the divided regions, and the computer Further, in each divided area, a step of counting the number of pixels in the specified pixel area, and a computer Determining whether the number of pixels in the specified pixel area is equal to or greater than a first threshold value in the divided area; and causing the computer to determine the number of pixels in the specified pixel area in the divided area. A step of calculating feature points of the pixel region and counting the number of feature points when determining that the number of feature points is equal to or greater than a first threshold in the divided region; And determining whether or not the number of feature points counted exceeds a second threshold value, the feature amount is calculated based on the feature points extracted from the pixel region. And causing the computer to execute a step of changing the first threshold value when it is determined that the counted number of feature points is equal to or less than the second threshold value. Characterized in that are recording the computer program.

  Further, the recording medium according to the present invention causes a computer to specify a plurality of pixel areas adjacent to pixels determined to have the same pixel value from a binary image, and based on the coordinate values of the pixels in the specified pixel area. A computer program for extracting feature points of the pixel region, calculating a feature amount indicating the feature of the image based on the extracted feature point, and determining similarity between the images based on the calculated feature amount; In a recording medium that can be read by a computer, the computer causes the computer to determine neighboring feature points located in the peripheral region of the extracted feature points, and causes the computer to count the number of the feature points in the determined neighborhood. A step, causing the computer to determine whether the score of the counted feature points is equal to or less than a second threshold; and causing the computer to count the feature points If it is determined that the score exceeds the second threshold, the step of calculating the feature amount based on the neighboring feature points and the computer determines that the score of the counted feature points is equal to or less than the second threshold In this case, a computer program for executing the step of changing the range of the peripheral area is recorded.

  Further, the recording medium according to the present invention causes a computer to specify a plurality of pixel areas adjacent to pixels determined to have the same pixel value from a binary image, and based on the coordinate values of the pixels in the specified pixel area. A computer program for extracting feature points of the pixel region, calculating a feature amount indicating the feature of the image based on the extracted feature point, and determining similarity between the images based on the calculated feature amount; In a recording medium that can be read by a computer, when extracting feature points, the computer counts the number of feature points within a predetermined range, and the computer counts the number of feature points into an area to be processed. A step of extracting feature points included, a step of causing a computer to determine whether or not the number of extracted feature points is smaller than a third threshold, If the data determines that the number of extracted feature points is equal to or greater than a third threshold, the step of calculating the feature amount based on neighboring feature points located in the peripheral region of the extracted feature points And a computer program for executing the step of changing the range of the peripheral area when it is determined that the extracted score is smaller than the third threshold.

  Further, the recording medium according to the present invention causes a computer to specify a plurality of pixel areas adjacent to pixels determined to have the same pixel value from a binary image, and based on the coordinate values of the pixels in the specified pixel area. A computer program for extracting feature points of the pixel region, calculating a feature amount indicating the feature of the image based on the extracted feature point, and determining similarity between the images based on the calculated feature amount; In a recording medium readable by a computer, a step of causing the computer to count the number of pixels in the specified pixel region, and whether the number of pixels in the specified pixel region is equal to or greater than a first threshold value Determining whether the number of pixels in the specified pixel area is greater than or equal to the first threshold, and extracting feature points of the pixel area A step of causing the computer to determine neighboring feature points located within the peripheral region of the extracted feature points, a step of causing the computer to count the number of feature points of the determined neighborhood, and A step of determining whether or not the number of feature points is equal to or less than a second threshold; and if the computer determines that the counted number of feature points exceeds the second threshold, based on the neighboring feature points Calculating the feature amount, and if the computer determines that the number of feature points counted is less than or equal to the second threshold value, the first threshold value, or the first threshold value and the surrounding region A computer program for executing the step of changing the range is recorded.

  Furthermore, the recording medium according to the present invention allows a computer to specify a plurality of adjacent pixel areas that are determined to have the same pixel value from a binary image, and to determine the coordinate value of the pixel of the specified pixel area. A computer program for extracting feature points of the pixel area based on the extracted feature points, calculating a feature amount indicating the feature of the image based on the extracted feature point, and determining similarity between the images based on the calculated feature amount; And having the computer count the number of pixels in the specified pixel area and whether the number of pixels in the specified pixel area is equal to or greater than a first threshold. When determining whether or not, a step of setting a plurality of different first threshold values and determining whether or not each of the first threshold values is equal to or greater than the first threshold value; and If it is determined that the threshold value is equal to or greater than the threshold value, the feature points of the pixel region are calculated and the number of feature points is counted for each of a plurality of different first threshold values. Extracting the feature points of the pixel area to be used, causing the computer to determine neighboring feature points located in the peripheral area of the extracted feature points, and causing the computer to count the number of feature points in the determined neighborhood A step of causing the computer to determine whether the score of the counted feature points is less than or equal to a second threshold; and a case of determining that the score of the counted feature points exceeds the second threshold Is a step of calculating the feature amount based on the nearby feature points, and if the computer determines that the number of feature points counted is equal to or less than a second threshold value. Characterized in that are recorded thereon a computer program for executing the steps of changing the range of the serial peripheral region.

  In the present invention, when the number of feature points is equal to or less than the second threshold, the feature points for calculating the feature amount are increased by changing the first threshold and increasing the number of feature points to be extracted from the pixel region. Adjustment can be made so that the score does not become extremely small, and the accuracy of similarity determination can be stabilized without deteriorating.

  Further, in the present invention, since feature points are extracted from pixel regions limited based on a plurality of different first threshold values, and feature quantities are calculated based on the largest number of extracted feature points, Adjustment can be made so that the number of feature points does not become extremely small, and it can be stabilized without reducing the accuracy of similarity determination.

  In the present invention, the document image is divided, and a lower limit value for the number of pixels in the connected region, that is, a first threshold value is set for each of the divided regions. It is possible not only to set a threshold value so that the noise in the upper, lower, left and right edge areas of the image is not calculated as the center of gravity, but also in the important area of the document image (for example, the center area) with respect to a smaller connected area. It is possible to set a threshold value for calculation, and a sufficient number of high-precision centroids can be secured.

  In the present invention, when extracting feature points, by counting the number of feature points in a predetermined range (pixel block line), the feature points are not read out every processing of the feature point of interest. Since the mask size or the number of reference blocks can be changed, it is not necessary to repeatedly extract neighboring feature points, and the processing can be facilitated and speeded up.

  Furthermore, in the present invention, a plurality of actions can be generated by combining these.

  In the present invention, it is possible to stably execute an image processing method that uses the center of gravity of a pixel region in an image, which is a similarity determination method for an image that is resistant to disturbances, without reducing the accuracy of similarity determination. This process is performed when performing a process for registering an image and a collation process between the input image and the registered image. When registering an image, for example, a registration mode may be selected from an operation panel of a digital copying machine (or a multifunction machine). When implemented as software, for example, a screen for setting the operation of the scanner may be displayed on the display, and the registration mode may be selected using a mouse or a keyboard.

  Hereinafter, an image processing method, an image processing apparatus, an image forming apparatus, a computer program, and a recording medium according to the present invention will be described with reference to the drawings illustrating embodiments. FIG. 1 is a schematic diagram showing an overall configuration of an image forming apparatus including an image processing apparatus 2 according to the present invention. Hereinafter, while substituting the main embodiment, the subordinate examples in each part of the description will be described as appropriate.

  The image forming apparatus in the figure is, for example, a digital copying machine, a multi-function printer (multifunction printer: MFP), and the like, and includes an image input device 1, an image processing device 2, an image output device 3, and an operation panel 4. . The image input device 1 is, for example, a scanner provided with a CCD (Charged Coupled Device). The image input device 1 receives reflected light from an original as an RGB analog signal and outputs the RGB analog signal to the image processing device 2. The image processing apparatus 2 performs processing described later on the received RGB analog signals and outputs the processed signals to the image output apparatus 3 as CMYK digital color signals. The image output device 3 is, for example, a color image output device using an electrophotographic method or an inkjet method, accepts CMYK digital color signals output from the image processing device 2, and uses paper or the like based on the received digital color signals A color image is output on the surface of the recording medium.

  The image processing apparatus 2 includes an A / D conversion unit 20, a shading correction unit 21, an image matching processing unit 22, an input tone correction unit 23, a region separation processing unit 24, a color correction unit 25, a black generation and under color removal unit 26, A spatial filter processing unit 27, an output tone correction unit 28, and a tone reproduction processing unit 29 are provided.

  The A / D converter 20 converts the RGB analog signal received from the image input device 1 into a digital signal, and outputs the converted digital signal to the shading correction unit 21.

  The shading correction unit 21 receives the digital signal output from the A / D conversion unit 20, and performs processing for removing various distortions generated in the illumination system, the imaging system, and the imaging system of the image input device 1 from the received digital signal. Do. In addition, the shading correction unit 21 converts the signal into an optimum signal for processing in the image processing apparatus 2, performs processing for adjusting the color balance, and outputs the processed digital signal to the image matching processing unit 22.

  The image collation processing unit 22 corresponds to the image processing apparatus according to the present invention, binarizes an image based on a digital signal of the received image, and generates a pixel region (connected component) in which pixels of the same value continue from the binary image. Based on the coordinate value of the pixel of the specified pixel region, the feature point of the pixel region is extracted, the feature amount indicating the feature of the image is calculated based on the extracted feature point, and based on the calculated feature amount The similarity determination between the images is performed, and a determination result indicating the result of the similarity determination is output to the memory 225 described later. A control unit 226, which will be described later, which is a CPU or MPU, performs predetermined processing according to the determination result. For example, when receiving a determination result indicating that the images are similar, the control unit 226 abandons output of the image, prohibits copying of the image, or stores the image in a predetermined folder (filing) )I do. In the present embodiment, the images are the images received via the image input device 1 to be determined this time, and the other document previously received via the image input device 1 and registered in a hash table described later. However, the present invention is not limited to this. The image matching processing unit 22 performs the above processing and outputs the digital signal received from the shading correction unit 21 to the input tone correction unit 23 without being modified.

  The input tone correction unit 23 receives the digital signal output from the image matching processing unit 22, performs image quality tone processing such as background density removal or contrast on the received digital signal, and processes the processed digital signal as a region. The data is output to the separation processing unit 24.

  The region separation processing unit 24 receives the digital signal output from the input tone correction unit 23, and distributes the image to any of a character region, a halftone region, and a photo region based on the received digital signal. The region separation processing unit 24 generates a region identification signal indicating whether the region belongs to a character region, a halftone dot region, or a photo region based on the sorting result, and the generated region identification signal is converted into a black generation and under color removal unit 26, In addition to outputting to the spatial filter processing unit 27, the output tone correction unit 28, and the tone reproduction processing unit 29, the digital signal received from the input tone correction unit 23 is output to the color correction unit 25.

  The color correction unit 25 receives the digital signal output from the region separation processing unit 24, performs a process of removing color turbidity due to the spectral characteristics of the CMY color material including unnecessary absorption components on the received digital signal, and has processed the digital signal. The signal is output to the black generation and under color removal unit 26.

  The black generation and under color removal unit 26 receives the digital signal (CMY) output from the color correction unit 25, generates a black signal (K) from the received digital signal, and subtracts the generated black signal from the digital signal. A new digital signal (CMY) is generated, and the generated new digital signal (CMY) and black signal (K) are output to the spatial filter processing unit 27, respectively.

  The spatial filter processing unit 27 receives the digital signal (CMYK) output from the black generation and under color removal unit 26 and performs a spatial filter process based on the region identification signal received from the region separation processing unit 24 on the received digital signal. The spatial frequency characteristic is corrected, and the processed digital signal is output to the output tone correction unit 28. Such processing can reduce blurring or graininess deterioration of an image to be output.

  The gradation reproduction processing unit 29 receives the digital signal output from the spatial filter processing unit 27, performs the following processing on the received digital signal, and stores the processed digital signal (CMYK) in a storage device (not shown). The image is stored, read at a predetermined timing, and output to the image output device 3. The gradation reproduction processing unit 29 selects a character region from the image based on the region identification signal received from the region separation processing unit 24, and emphasizes the high-frequency component of the selected character region using a digital filter. Such processing can improve the reproducibility of characters. Further, the gradation reproduction processing unit 29 performs binarization or multi-value processing of an image using a high-resolution screen that is optimal for reproducing high-frequency components. The gradation reproduction processing unit 29 selects a halftone dot region from the image based on the region identification signal, and removes an input halftone dot component from the selected halftone dot region using a low-pass filter. The gradation reproduction processing unit 29 selects a photographic region from the image based on the region identification signal, and performs binarization or multi-value processing on the selected photographic region with a screen that emphasizes gradation reproducibility.

  Next, the image matching processing unit 22 will be described. FIG. 2 is a schematic diagram showing the overall configuration of the image matching processing unit 22. The image matching processing unit 22 includes a feature point calculation unit 221, a feature amount calculation unit 222, a voting processing unit 223, a similarity determination processing unit 224, a memory 225, a control unit 226, and a registration processing unit 227.

  The feature point calculation unit 221 specifies a plurality of connected components in which pixels having the same value continue from the received binary image data, extracts feature points of the connected components based on the coordinate values of the pixels of the specified connected components, and extracts them The obtained feature points are output to the feature amount calculation unit 222. The configuration and processing procedure of the feature point calculation unit 221 will be described later.

  The feature amount calculation unit 222 calculates a feature amount indicating the feature of the image based on the feature points calculated by the feature point calculation unit 221. The configuration and processing procedure of the feature amount calculation unit 222 will be described later.

  The voting processing unit 223 searches the hash table stored in the memory 225 based on the feature amount (hash value) calculated by the feature amount calculation unit 222, and vote for the index image registered in the hash table. And the result of cumulative addition of the votes is output to the similarity determination processing unit 224.

  The similarity determination processing unit 224 receives the result output from the voting processing unit 223, determines image similarity based on the received result, and outputs the determination result of the determination via the registration processing unit 227. At this time, the processing in the registration processing unit 227 is through (no processing is performed). A specific processing procedure will be described later.

  The memory 225 stores data generated by processing executed by the image matching processing unit 22, a hash table described later, and the like.

  The control unit 226 controls the hardware included in the image collation processing unit 22 to perform a series of processes.

  The registration processing unit 227 sequentially registers the feature amount (hash value) calculated by the feature amount calculation unit 222 and the index (document ID) representing the document (registered image) in the hash table stored in the memory 225. (See FIG. 23 (a)). If the hash value is already registered, the document ID is registered in association with the hash value.

  Next, the feature point calculation unit 221 of the image matching processing unit 22 will be described. FIG. 3 is a schematic diagram illustrating the configuration of the feature point calculation unit 221. The feature point calculation unit 221 includes an achromatic processing unit 2210, a resolution conversion unit 2211, an MTF correction processing unit 2212, a binarization processing unit 2213, and a centroid calculation unit 2214.

When the received image data is a color image, the achromatic processing unit 2210 performs processing for achromatizing the image data and converting it to a brightness or luminance signal. For example, the achromatic processing unit 2210 calculates Y _j = 0.30R _j + 0.59G _j + 0.11B _j and performs achromatic processing using the calculation result. Where Y _{j is} the luminance value of each pixel, and R _j G _j B _j is the color component of each pixel. The RGB signal may be converted into a CIE 1976 L * a * b * signal. However, CIE: Commission International de l'Eclairage, L *: lightness, a * b *: chromaticity.

  The resolution conversion unit 2211 performs a process of scaling again so as to obtain a predetermined resolution when the received image data is optically scaled by the image input apparatus 1. When the image data is not optically scaled, the resolution conversion unit 2211 can reduce the capacity to be processed by converting the image data to a low resolution so as to reduce the processing.

  The MTF correction processing unit 2212 performs an appropriate filter process on the image data received from the image input device 1 having different spatial frequency characteristics for each model, thereby degrading the image data based on the spatial frequency characteristics of the image data. To repair. In addition, the MTF correction processing unit 2212 performs enhancement and smoothing processing using a mixing filter in order to suppress unnecessary high-frequency components.

  The binarization processing unit 2213 converts the received image data that is achromatic or the image data achromatic by the achromatization processing unit 2210 into binary image data suitable for calculating the center of gravity described later. I do. For example, the binarization processing unit 2213 calculates an average value of the pixel values of the image, and performs a process of binarizing each pixel of the image using the calculated average value as a threshold value.

  The center-of-gravity calculation unit 2214 obtains the center of gravity of a connected component in which pixels having the same value continue from the binary image data binarized by the binarization processing unit 2213 by processing described later, and outputs the center of gravity to the memory 225 as the feature point. To do. FIG. 4 is a schematic diagram showing the configuration of the center-of-gravity calculation unit 2214. The centroid calculation unit 2214 includes a labeling processing unit 2214a, a connected component threshold processing unit 2214b, a centroid calculation processing unit 2214c, a centroid storage buffer 2214d, and a control unit 2214e.

  The labeling processing unit 2214a identifies the connected component from the image and sequentially labels the pixels of the identified connected component.

  The control unit 2214e cumulatively adds the coordinate values of the labeled pixels, and stores the addition result in the centroid storage buffer 2214d.

  The center-of-gravity storage buffer 2214d stores an x coordinate addition result, a y coordinate addition result, the number of times of addition to each label (that is, the number of labeled pixels), and a flag used for determination.

  The control unit 2214e ends the cumulative addition process when there is a non-target label after the cumulative addition for one line of the image is completed, that is, when it is not continuous with the pixel of the next line. In addition, the control unit 2214e changes the first threshold value.

  The connected component threshold processing unit 2214b performs a centroid calculation determination described later based on the addition result stored in the centroid storage buffer 2214d.

  When the connected component threshold processing unit 2214b makes a centroid calculation determination, the centroid calculation processing unit 2214c performs centroid calculation based on the addition result stored in the centroid storage buffer 2214d, and uses the calculated centroid as a feature point to the memory 225. Output.

  Next, the procedure of the centroid calculation process performed by the centroid calculation unit 2214 of the feature point calculation unit 221 will be described. FIG. 5 is a flowchart showing the procedure of the centroid calculation process of the centroid calculation unit 2214. Note that the variables shown in the drawing correspond to the addition result of the x coordinate, the addition result of the y coordinate, the number of times of addition to each label (that is, the number of labeled pixels), and a flag used for determination.

  cnt is an internal counter indicating the maximum value of label assignment for the connected component. For example, when a maximum of 2047 labels is given, cnt is represented by 11 bits (0 to 2047). That is, when cnt = 11, it is indicated that at least the number of identified connected components is 11.

  Further, renew_flg is a 2-bit signal flag indicating whether or not the corresponding label has been updated, and is represented as “00b” when the corresponding label information is empty, and “01b” when the corresponding label information is not updated. When the information of the corresponding label is updated, it is expressed as “10b”. For example, renew_flg [cnt] is an update flag when data having a cnt value as an address is read from the centroid storage buffer 2214d, and renew_flg [5] represents an update flag of data stored at address 5.

  Further, lbnum indicates the number of pixels of the corresponding label, and 1 is incremented to the lbnum value of the corresponding label every time labeling is performed. For example, lbnum [cnt] represents the number of pixels of the label when data having the cnt value as an address is read from the centroid storage buffer 2214d. As a result, the number of connected component pixels can be counted.

  Also, sumx indicates the sum obtained by adding the x-coordinates of the pixels of the corresponding label, and the x-coordinate of the current target pixel is added to the sumx value of the corresponding label every time labeling is performed. For example, sumx [cnt] represents the x coordinate sum when data having the cnt value as an address is read from the centroid storage buffer 2214d.

  Further, sumy indicates the sum obtained by adding the y-coordinates of the pixels of the corresponding label, and the y-coordinate of the current target pixel is added to the sumy value of the corresponding label every time labeling is performed. For example, sumy [cnt] represents the y-coordinate sum when data having the cnt value as an address is read from the centroid storage buffer 2214d.

  Further, lbtbl indicates label information when the label table is read, and lbtbl [cnt] indicates label information when the label table with the cnt value as an address is read from the centroid storage buffer 2214d.

  When the process is started, the center-of-gravity calculation unit 2214 initializes cnt by the labeling processing unit 2214a (S101), determines whether or not the current cnt is 0 (S102), and if cnt is 0 (S102) YES), the process ends. The initial value is equal to the specified number of connected components, and is either 2047 which is the maximum number of labels or the maximum value of cnt.

  On the other hand, when the labeling processing unit 2214a determines that cnt is not 0 (NO in S102), the center-of-gravity calculation unit 2214 reads data having the current cnt value as an address from the center-of-gravity storage buffer 2214d (S103). It is determined whether or not renew_flg [cnt] included in is 00b (S104). That is, the center-of-gravity calculation unit 2214 determines whether or not the information on the corresponding label is empty. As a result, if the labeling processing unit 2214a determines that renew_flg [cnt] is 00b (YES in S104), the center-of-gravity calculation unit 2214 updates an empty flag indicating that the corresponding label is empty (S105). The current cnt is decremented (S106), and the process returns to the determination in step S102.

  On the other hand, if the labeling processing unit 2214a determines that renew_flg [cnt] is not 00b (NO in S104), the center-of-gravity calculation unit 2214 further determines whether or not lbtbl [cnt] is equal to the current cnt value (S107). ), When it is determined that lbtbl [cnt] is not equal to the current cnt value (NO in S107), data having the lbtbl [cnt] value as an address is read from the centroid storage buffer 2214d (S108).

  The center-of-gravity calculation unit 2214 performs addition processing, which will be described later, on the data read by the labeling processing unit 2214a (S109), and updates the data with the lbtbl [cnt] value as an address as a result of the addition processing (S110).

  The center-of-gravity calculation unit 2214 initializes the data read by the labeling processing unit 2214a using the cnt value as an address (S111), decrements the current cnt (S106), and returns to the determination in step S102.

  On the other hand, if the labeling processing unit 2214a determines that lbtbl [cnt] is equal to the current cnt value (YES in S107), the center-of-gravity calculation unit 2214 further determines whether or not renew_flg [cnt] is 01b (S112). ). That is, it is determined whether or not the information on the corresponding label has been updated. As a result, if the labeling processing unit 2214a determines that renew_flg [cnt] is 01b (YES in S112), the center-of-gravity calculation unit 2214 executes threshold determination processing described later (S113), and uses the cnt value as an address. The read data is initialized (S114), the empty flag indicating that the corresponding label is empty is updated (S115), the current cnt is decremented (S106), and the process returns to the determination in step S102.

  On the other hand, if the labeling processing unit 2214a determines that renew_flg [cnt] is not 01b (NO in S112), the center-of-gravity calculation unit 2214 updates renew_flg [cnt] to 01b (S116), that is, the information of the corresponding label is It shows that it was reset without updating, decrements the current cnt (S106), and returns to the determination in step S102.

  Next, the threshold determination procedure executed by the gravity center calculation unit 2214 in step S113 of the gravity center calculation process will be described. The center-of-gravity calculation unit 2214 counts the number of pixels lbnum of the target label, determines whether or not the counted number of pixels lbnum of the target label is equal to or greater than a first threshold value, and performs center-of-gravity calculation processing according to the determination result It is determined whether or not to calculate the center of gravity in the part 2214c. FIG. 6 is a flowchart showing the threshold determination procedure of the center-of-gravity calculation unit 2214.

  The center-of-gravity calculation unit 2214 reads the cnt value as an address by the connected component threshold processing unit 2214b, and determines whether or not lbnum [cnt], which is the number of pixels of the target label counted, is less than the first threshold (S201). ), If it is determined that lbnum [cnt] is less than the first threshold (YES in S201), the data including lbnum [cnt] is discarded (S202), and the first threshold is changed as described later. Then, the threshold determination is finished.

  On the other hand, if the connected component threshold processing unit 2214b determines that lbnum [cnt] is equal to or greater than the first threshold (NO in S201), the center-of-gravity calculation unit 2214 calculates sumx, sumy, and lbnum included in the read data. The data is output to the processing unit 2214c (S203), and the threshold determination ends.

  Here, a specific example of the threshold determination will be shown. FIG. 8 is an explanatory diagram showing an example of threshold determination, and FIG. 9 is an explanatory diagram showing an example of a document with few feature points. For example, when the first threshold value is set to 100, the connected component of the label “I” in FIG. 8 is the center of gravity calculation target because the number of pixels of the left connected component is 109. Since the number of pixels of the right connected component is 38, the center of gravity calculation is not performed and the obtained result is discarded. Therefore, when the document including the character “I” in FIG. 8 is composed of a ruled line frame and a small number of characters (see FIG. 9), the number of pixels of the connected component is small, the center of gravity calculation is reduced, and the feature points The required number of points cannot be secured, and the determination accuracy is lowered. Therefore, the center-of-gravity calculation unit 2214 changes the first threshold value to 30, for example. As a result, the right connected component becomes an object of the center of gravity, and the number of feature points increases, so that it is possible to avoid a decrease in determination accuracy.

  Next, the procedure of the addition process executed by the gravity center calculation unit 2214 in step S109 of the gravity center calculation process will be described. The center-of-gravity calculation unit 2214 cumulatively adds the coordinate values of the pixels of the connected component of the label, counts the number of pixels used for the cumulative addition, and outputs these results to the center-of-gravity calculation processing unit 2214c. FIG. 7 is a flowchart showing the procedure of the addition process of the centroid calculating unit 2214.

  The center-of-gravity calculation unit 2214 adds sumx included in the data read out from the center-of-gravity storage buffer 2214d as the address by the control unit 2214e to sumx included in the data read out from the buffer as the lbtbl [cnt] value (S301). Then, the sumy included in the data read out from the buffer using the cnt value as an address is added to the sumy included in the data read out from the buffer using the lbtbl [cnt] value as an address (S302). The center-of-gravity calculation unit 2214 adds the lbnum included in the data read from the buffer with the cnt value as the address by the control unit 2214e to the lbnum included in the data read out from the buffer as the lbbtbl [cnt] value (S303). It is determined whether or not the subsequent lbnum exceeds the maximum value FFFh (S304). As a result, when the control unit 2214e determines that the added lbnum exceeds FFFh (YES in S304), the center-of-gravity calculation unit 2214 selects FFFh (S305) and clips the selected FFFh.

  On the other hand, when the control unit 2214e determines that the added lbnum does not exceed FFFh (NO in S304), the center-of-gravity calculation unit 2214 selects the added lbnum (S306) and clips the selected added lbnum. To do.

  Then, the center-of-gravity calculation unit 2214 determines whether or not renew_flg [cnt] is 10b in the control unit 2214e (S307), and determines that renew_flg [cnt] is 10b (YES in S307). Renew_flg [lbtbl [cnt]] included in the data read out from the buffer 2214d using the lbtbl [cnt] value as an address is updated to 10b (S309), and the addition process is terminated.

  On the other hand, if the control unit 2214e determines that renew_flg [cnt] is not 10b (NO in S307), the center-of-gravity calculation unit 2214 does not update renew_flg [lbtbl [cnt]] (S308) and ends the addition process.

  After the addition process is completed, the center-of-gravity calculation unit 2214 updates the data read by the control unit 2214e using the lbbtbl [cnt] value as an address to sumx and sumy after addition and lbnum or FFFh after addition (step S110). reference).

  In the present embodiment, the center-of-gravity calculation unit 2214 performs threshold determination based on one first threshold. However, the present invention is not limited to this, and threshold determination is performed in parallel based on a plurality of different first thresholds. As a result of the determination, a first threshold value that maximizes the number of connected components may be determined, and the center-of-gravity calculation process, threshold value determination, and addition process may be executed based on the determined first threshold value. This will be described as Example 1.

Example 1.
FIG. 10 is a schematic diagram illustrating a configuration of the gravity center calculation unit 2214 according to the first embodiment. In Example 1, connected component threshold processing section 2214b includes a first connected component threshold processing section 2214b ₁ and the second connected component threshold processing section 2214b _2, centroid calculation processing section 2214C first centroid calculation processing section 2214C ₁ and the second The center-of-gravity calculation processing unit 2214c ₂ is used, and the center-of-gravity storage buffer 2214d is a first center-of-gravity storage buffer 2214d ₁ and a second center-of-gravity storage buffer 2214d ₂ . The center-of-gravity calculation unit 2214 according to another embodiment determines a threshold based on a first threshold (for example, 30, 100) that is different from the first connected component threshold and the second connected component threshold, and calculates the center of gravity based on the determination result. The selection unit 2215f selects the result of the largest center of gravity (feature point), and outputs the selected feature point to the memory 225. The first embodiment has been described above.

  In the present embodiment, an example in which the lower limit value, that is, the first threshold value is changed without dividing the document image has been described. However, the present invention is not limited to this, and the document image is divided and the divided area is divided. You may make it change a 1st threshold value in each. Hereinafter, the second embodiment will be described.

Example 2
FIG. 28 is an explanatory diagram showing an example of an input document image. There are a wide variety of document formats that are generally input, but there are few cases where characters exist uniformly throughout the entire document, and there are many cases where characters do not exist in the upper, lower, left, and right areas of the document. . For example, as shown in FIG. 28, when the lower limit of threshold determination is uniformly set to 100, noise falls within the threshold determination range, and thus becomes a target of gravity center calculation. Further, when the number of pixels of the connected component of the character existing in the central area of the document image is 100 or less (for example, the number of pixels at the point i), the center of gravity is not calculated. From the above, the accuracy of the calculated center of gravity may be reduced.

  Therefore, in the second embodiment, a threshold value is set so that noise in the upper, lower, left, and right end regions is not calculated as the center of gravity by dividing the document image and setting a lower limit value for each of the divided regions. In addition, it is possible to set a threshold value for calculating the center of gravity for a smaller connected area even in an important area (for example, the central area) of the document image. Flexible threshold setting aiming at accuracy improvement is possible. That is, since the centroid can be calculated for the connected component with a small character portion as well as the noise is removed, a sufficient number of centroids with high accuracy can be secured.

  FIG. 29 is an explanatory diagram illustrating an example in which the document image is equally divided into four parts, FIG. 30 is a schematic diagram illustrating a configuration of the centroid calculating unit 2214 according to the second embodiment, and FIGS. 31 and 32 are explanatory diagrams illustrating the structure of the threshold value table. FIG. 33 is an explanatory diagram showing an example of dividing a document image.

  In FIG. 29, areas 1, 2, 3, and 4 are defined in order from the upper end area. In the second embodiment, the lower limit threshold of the region 1 is set to 200, the lower limit threshold of the region 2 is set to 30, the lower limit threshold of the region 3 is set to 30, and the lower limit threshold of the region 4 is set to 200. With this configuration, in the region 1, only the character “A” having 400 pixels is a centroid calculation target, and noise is not a target. In the region 2 and the region 3, a character connected component having a small number of pixels such as a point i is also a centroid calculation target. In the region 4, only noise is present, but this noise is not a centroid calculation target.

  From the above, the center of gravity can be calculated with high accuracy by the method of the second embodiment. Further, according to this method, it is not necessary to take a procedure of performing threshold determination in parallel and determining a range from the result of the determination, so that the processing can be facilitated and performed at high speed.

  The center-of-gravity calculation unit 2214 includes a threshold value table (see FIG. 30). As shown in FIG. 31, the threshold value table may store the lower limit threshold value by associating the address with the area number, and as shown in FIG. 32, the address is associated with the division number of the area of the document image. A configuration in which each lower limit threshold value is stored may be employed. The other configuration of the centroid calculating unit 2214 is the same as the configuration of the centroid calculating unit 2214 shown in FIG.

  Regarding the number of divisions of the document image, the number of divisions may be determined by taking statistics from various document images. As another method, the number of divisions may be determined using the type of document image or the result of layout recognition.

  As shown in FIG. 29, regarding the division of the document image, it is not necessary to uniformly divide the image according to the number of divisions, and the number of lines may be set for each region, or automatically calculated from the set weights. For example, a configuration in which a weighting factor is set and calculated may be used. Further, as shown in FIG. 33, the main scanning direction and sub-scanning direction offset (HOFT, VOFT) and the number of pixels (HMAX, VMAX) are not divided into only one of the main scanning direction and the sub-scanning direction. ), The document image may be divided into upper and lower left and right regions and a central region, and different lower threshold values may be set. In the example of FIG. 33, the number of divisions is 2.

  When the number of feature points calculated by dividing the document image described above and setting the lower limit value of the connected region in each divided region is equal to or smaller than the second threshold value, The range (mask size, number of reference blocks) may be changed. The second embodiment has been described above.

  Next, the feature amount calculation unit 222 of the image matching processing unit 22 will be described. FIG. 11 is a schematic diagram illustrating a configuration of the feature amount calculation unit 222, and FIG. 12 is an explanatory diagram illustrating a peripheral region of the feature point of interest. The feature amount calculation unit 222 includes a pixel block setting unit 2220, a neighboring point extraction unit 2221, and a feature amount extraction unit 2222.

  The pixel block setting unit 2220 sets any one of the extracted feature points as a target feature point, sets a pixel block including one or a plurality of pixels of the image to indicate the peripheral region of the target feature point, and sets the peripheral region To change the mask size of the set pixel block or the number of reference blocks.

  The neighboring point extraction unit 2221 determines, among the extracted feature points, neighboring feature points located in the peripheral region, counts the number of neighboring feature points determined, and the counted number is equal to or less than a second threshold value. The process which determines whether it is is is performed. As a result of the determination, when it is determined that the counted score exceeds the second threshold, the feature amount extraction unit 2222 performs a feature amount calculation process. If there are many feature points in the vicinity located in the peripheral region, four feature points that are at the shortest distance from the feature point of interest may be determined as neighboring feature points (see FIG. 12). .

  The feature amount extraction unit 2222 performs a process of calculating a feature amount indicating the feature of the image based on the target feature point and / or the determined nearby feature point, and outputs the calculated feature amount to the memory 225. A method for calculating the feature amount will be described later.

  Next, the relationship between the number of neighboring feature points located in the peripheral area of the feature point of interest and the change of the peripheral area will be described. FIG. 13 is an explanatory diagram showing the relationship between the mask size of the peripheral area and the neighboring feature points located in the peripheral area.

  For example, when one of the feature points extracted by the pixel block setting unit 2220 is set as the feature point of interest, the feature amount calculation unit 222 selects the pixel block so that the mask size is 256 pixels × 256 pixels per one. Then, a neighboring 8 pixel block centered on the pixel block including the feature point of interest is set as a “peripheral region”. At this time, the feature quantity calculation unit 222 can extract three neighboring feature points located in the peripheral region by the neighboring point extraction unit 2221 (see FIG. 13A). Therefore, the feature amount calculation unit 222 changes the pixel block mask size so that the pixel block setting unit 2220 has 512 pixels × 512 pixels per pixel. The feature amount calculation unit 222 can extract four feature points in the vicinity located in the peripheral region by the vicinity point extraction unit 2221 (see FIG. 13B). In other words, it is possible to avoid a decrease in determination accuracy by increasing the number of feature points.

  Here, the surrounding area setting process, the neighboring feature point determining process, and the feature quantity calculating process of the feature quantity calculating unit 222 will be described. FIG. 14 is a flowchart illustrating a processing procedure of the feature amount calculation unit 222.

  The feature amount calculation unit 222 sets a pixel block including one or a plurality of pixels of the image in the pixel block setting unit 2220, and sets the mask size of the set pixel block (S401). The feature quantity calculation unit 222 sets any one of the feature points extracted by the pixel block setting unit 2220 as a target feature point, and sets the neighboring eight pixel blocks centered on the pixel block including the target feature point as a “peripheral region”. The data including the set pixel block, mask size, and peripheral region is output to the neighborhood point extraction unit 2221.

  The feature quantity calculation unit 222 receives the data output from the pixel block setting unit 2220 by the neighboring point extraction unit 2221 and extracts neighboring feature points located in the peripheral area based on the received data (S402). The feature quantity calculation unit 222 counts the number of neighboring feature points extracted by the neighboring point extraction unit 2221 (S403), and determines whether the counted number is equal to or less than a preset second threshold value (S404). ). As a result, when it is determined that the score counted by the neighboring point extraction unit 2221 is equal to or less than the second threshold (YES in S404), the feature amount calculation unit 222 outputs a signal indicating that to the pixel block setting unit 2220. To do. The feature amount calculation unit 222 changes the mask size of the pixel block by the pixel block setting unit 2220 that has received the signal (S405), and outputs data including the changed mask size to the neighborhood point extraction unit 2221. The feature amount calculation unit 222 repeats Step S402 in the neighborhood point extraction unit 2221.

  On the other hand, when it is determined that the number of points counted by the neighboring point extraction unit 2221 exceeds the second threshold (NO in S404), the feature amount calculation unit 222 sends the data related to the extracted neighboring feature points to the feature amount extraction unit 2222. Output. The feature amount calculation unit 222 receives the data output from the neighboring point extraction unit 2221 by the feature amount extraction unit 2222, and calculates the feature amount based on the received data (S406). The feature amount calculation unit 222 outputs the feature amount calculated by the feature amount extraction unit 2222 to the memory 225, and ends the process.

  In the present embodiment, the feature quantity calculation unit 222 performs the feature quantity calculation by the feature quantity extraction unit 2222 when certain conditions are satisfied (see steps S404 and S406). The calculation unit 222 may determine a certain condition after the feature amount extraction unit 2222 executes the feature amount calculation. Hereinafter, the third embodiment will be described.

Example 3 FIG.
FIG. 15 is a flowchart illustrating a processing procedure of the feature amount calculation unit 222 according to the third embodiment. The feature amount calculation unit 222 according to the third embodiment sets a pixel block including one or a plurality of pixels in the image by the pixel block setting unit 2220, and sets a mask size of the set pixel block (S501). The feature quantity calculation unit 222 sets any one of the feature points extracted by the pixel block setting unit 2220 as a target feature point, and sets the neighboring eight pixel blocks centered on the pixel block including the target feature point as a “peripheral region”. The data including the set pixel block, mask size, and peripheral region is output to the neighborhood point extraction unit 2221.

  The feature amount calculation unit 222 receives the data output from the pixel block setting unit 2220 by the neighboring point extraction unit 2221 and extracts neighboring feature points located in the peripheral region based on the received data (S502). The feature amount calculation unit 222 outputs data related to the nearby feature points extracted by the neighborhood point extraction unit 2221 to the feature amount extraction unit 2222. The feature amount calculation unit 222 receives the data output from the neighboring point extraction unit 2221 by the feature amount extraction unit 2222, calculates the feature amount based on the received data (S503), and the calculated feature amount is valid. It is determined whether or not (S504). As a result, when the feature amount calculation unit 222 determines that the feature amount calculated by the feature amount extraction unit 2222 is not valid (NO in S504), the feature amount calculation unit 222 outputs a signal indicating that to the pixel block setting unit 2220. The feature amount calculation unit 222 changes the mask size of the pixel block by the pixel block setting unit 2220 that has received the signal (S505), and outputs data including the changed mask size to the neighborhood point extraction unit 2221. The feature amount calculation unit 222 performs the process of step S502 at the neighboring point extraction unit 2221.

  On the other hand, if the feature amount calculation unit 222 determines that the feature amount calculated by the feature amount extraction unit 2222 is valid (YES in S504), the feature amount calculation unit 222 outputs the calculated feature amount to the memory 225 and ends the process.

  Here, the determination of whether or not the calculated feature amount is valid is, for example, that only two points can be secured when the number of feature points to be extracted in step S503 is 4, and the feature is based on the two feature points. When the amount is calculated (S503), the maximum coordinate value is given as an initial value to the two points that could not be secured, and it is determined whether or not a distance equal to or larger than the diagonal distance of the mask size is calculated. The example 3 has been described above.

  Further, in this embodiment, an example of changing the mask size without counting the centroid in units of pixel block lines has been described. However, the present invention is not limited to this, and the centroid is counted in units of pixel block lines. The mask size may be changed according to the above. Hereinafter, this will be described as a fourth embodiment.

Example 4
FIG. 34 is a schematic diagram showing the configuration of the centroid calculating unit 2214 for counting the centroid in units of pixel block lines. The center of gravity calculation unit 2214 includes a center of gravity count buffer 2214f. The centroid count buffer 2214f stores the number of centroids counted in units of pixel block lines, and is configured to add 1 to the counter corresponding to the corresponding pixel block when storing the centroid. The other configuration of the centroid calculating unit 2214 is the same as the configuration of the centroid calculating unit 2214 shown in FIG.

  FIG. 35 is an explanatory diagram showing an example of counting by the centroid count buffer 2214f. FIG. 35 shows the count result of the pixel block line 1 having a count result of 14, the pixel block line 2 having a count result of 12, and thereafter the pixel block line 32. These count results are stored in the center-of-gravity count buffer 2214f.

  FIG. 36 is a flowchart illustrating a processing procedure of the feature amount calculation unit 222 according to the fourth embodiment. The feature amount calculation unit 222 first sets a mask size corresponding to the pixel block stored in the centroid count buffer 2214f (S801).

  The feature amount calculation unit 222 reads out the centroid count result corresponding to the pixel block line to be processed including the pixel block including the target point and the surrounding blocks from the centroid count buffer 2214f (S802), That is, the barycentric count results of neighboring mask lines are added (S803).

  The feature amount calculation unit 222 determines whether or not the addition result is smaller than the third threshold value (S804). Here, when extracting feature points of four neighboring points, it is necessary to change the mask size because the feature amount cannot be accurately calculated unless there are four or more feature points around the target feature point.

  When it is determined that the addition result is smaller than the third threshold value (YES in S804), the feature amount calculation unit 222 changes the mask size to a larger mask size because there is no neighboring feature point necessary for calculating the feature point. (S805), the process returns to step S802, and data is read from the centroid count buffer 2214f.

  On the other hand, when it is determined that the addition result is equal to or greater than the third threshold value (NO in S804), the feature amount calculation unit 222 extracts nearby feature points (S806) and outputs the feature points to the feature amount extraction unit 2222. The feature amount extraction unit 2222 calculates a feature amount based on the received data (S807). The feature amount calculation unit 222 ends the process.

  FIG. 37 is a schematic diagram for explaining threshold determination and mask size change. In FIG. 37, the mask size setting is set to 256 pixels × 256 pixels. In this case, the count number of the pixel block line including the feature point of interest BCCNT3 = 1, and the count numbers of the pixel block lines including the peripheral blocks are BCCNT2 = 1 and BCCNT4 = 2, respectively, so that the addition result is 4. The third threshold value (TH_SPREAD) is 4 neighboring feature points + one feature point of interest = 5. Accordingly, the feature amount calculation unit 222 determines that the addition result is smaller than the third threshold value (refer to YES in S804), and changes the initially set mask size of 256 pixels × 256 pixels to 512 pixels × 512 pixels. To do. Thereby, four neighborhood feature points can be secured.

  Note that the document image described above is divided, a lower limit value of a connected area is set in each divided area, and a process including a pixel block including a point of interest and its surrounding blocks is set in each divided area. The centroid count result corresponding to the pixel block line to be performed may be read from the centroid count buffer 2214f and added, and the mask size may be changed when the addition result is equal to or smaller than the second threshold value.

  When the centroid count result corresponding to the pixel block including the target point and the pixel block line that performs processing including the surrounding blocks is read from the centroid count buffer 2214f and added, and the addition result is smaller than the third threshold value The lower limit value of the number of pixels from which a connected area for obtaining a feature point is extracted or the lower limit value of the number of pixels from which a connected area for obtaining a feature point is extracted and the mask size may be changed.

  Furthermore, a plurality of lower limit values of the number of pixels for extracting a connected region for obtaining a feature point may be set, and the mask size may be changed when the most counted feature points are smaller than the third threshold value. In the above, Example 4 was demonstrated.

  Next, the relationship between the number of feature points in the vicinity located in the peripheral region of the feature point of interest and the change of the peripheral region will be described based on the third embodiment. FIG. 16 is an explanatory diagram showing the relationship between the number of pixel blocks in the peripheral area and the nearby feature points located in the peripheral area.

  For example, the feature amount calculation unit 222 sets the pixel blocks so that the number of reference blocks is 3 × 3 per one when any one of the feature points extracted by the pixel block setting unit 2220 is the target feature point. Then, a neighboring pixel block centering on the pixel block including the feature point of interest is set as a “peripheral region”. At this time, the feature quantity calculation unit 222 can extract three neighboring feature points located in the peripheral region by the neighboring point extraction unit 2221 (see FIG. 13A). Therefore, the feature amount calculation unit 222 changes the number of reference blocks of the pixel block at the pixel block setting unit 2220 so as to be 5 × 5. As a result, the feature amount calculation unit 222 can extract four feature points in the vicinity located in the peripheral region by the neighboring point extraction unit 2221 (see FIG. 16), and the number of feature points increases. A decrease in accuracy can be avoided.

  Here, the surrounding area setting process, the neighboring feature point determining process, and the feature quantity calculating process of the feature quantity calculating unit 222 will be described. FIG. 17 is a flowchart illustrating a processing procedure of the feature amount calculation unit 222 according to the third embodiment.

  The feature amount calculation unit 222 sets a pixel block including one or a plurality of pixels in the image by the pixel block setting unit 2220, and sets the reference block number of the set pixel block (S601). The feature amount calculation unit 222 sets any one of the feature points extracted by the pixel block setting unit 2220 as a target feature point, and sets a neighboring pixel block centered on the pixel block including the target feature point as a “peripheral region”. Then, the data including the set pixel block, the number of reference blocks, and the peripheral region is output to the neighborhood point extraction unit 2221.

  The feature amount calculation unit 222 receives the data output from the pixel block setting unit 2220 by the neighboring point extraction unit 2221, extracts neighboring feature points located in the surrounding area based on the received data (S602), and extracts them. The number of feature points in the vicinity is counted (S603), and it is determined whether or not the counted number is equal to or less than a preset second threshold value (S604). As a result, when it is determined that the score counted by the neighboring point extraction unit 2221 is equal to or less than the second threshold (YES in S604), the feature amount calculation unit 222 outputs a signal indicating that to the pixel block setting unit 2220. To do. The feature amount calculation unit 222 changes the reference block number of the pixel block by the pixel block setting unit 2220 that has received the signal (S605), and outputs data including the changed reference block number to the neighborhood extraction unit 2220. The feature amount calculation unit 222 performs step S602 in the neighborhood point extraction unit 2221.

  On the other hand, if it is determined that the number of points counted by the neighborhood point extraction unit 2221 exceeds the second threshold (NO in S604), the feature amount calculation unit 222 sends the data related to the extracted nearby feature points to the feature amount extraction unit 2222. Output. The feature amount calculation unit 222 receives the data output from the neighboring point extraction unit 2221 by the feature amount extraction unit 2222, calculates the feature amount based on the received data (S606), and outputs the calculated feature amount to the memory 225. Then, the process ends. In the above, it demonstrated based on Example 3. FIG.

  In the description based on the third embodiment, the feature amount calculation unit 222 has shown an example in which the feature amount extraction unit 2222 executes the feature amount calculation when a certain condition is satisfied (see steps S604 and S606). Not limited to this, after the feature amount extraction unit 2222 executes the feature amount calculation, the feature amount calculation unit 222 may determine certain conditions. Hereinafter, this will be described as a fifth embodiment.

Example 5 FIG.
FIG. 18 is a flowchart illustrating a processing procedure of the feature amount calculation unit 222 according to the fifth embodiment. The feature amount calculation unit 222 sets a pixel block including one or a plurality of pixels in the image by the pixel block setting unit 2220, and sets the reference block number of the set pixel block (S701). The feature quantity calculation unit 222 uses any one of the feature points extracted by the pixel block setting unit 2220 as a target feature point, and sets a neighboring pixel block centered on the pixel block including the target feature point as a “peripheral region”. The data including the set pixel block, the number of reference blocks, and the surrounding area is output to the neighborhood point extraction unit 2221.

  The feature amount calculation unit 222 receives the data output from the pixel block setting unit 2220 by the neighboring point extraction unit 2221 and extracts neighboring feature points located in the peripheral region based on the received data (S702). The feature amount calculation unit 222 outputs data related to the nearby feature points extracted by the neighborhood point extraction unit 2221 to the feature amount extraction unit 2222. The feature amount calculation unit 222 receives the data output from the neighboring point extraction unit 2221 in the feature amount extraction unit 2222, calculates the feature amount based on the received data (S703), and whether the calculated feature amount is valid. It is determined whether or not (S704). As a result, when it is determined that the feature amount calculated by the feature amount extraction unit 2222 is not valid (NO in S704), the feature amount calculation unit 222 outputs a signal indicating that to the pixel block setting unit 2220. The feature amount calculation unit 222 changes the reference block number of the pixel block by the pixel block setting unit 2220 that has received the signal (S705), and outputs data including the changed reference block number to the neighboring point extraction unit 2221. The feature amount calculation unit 222 performs step S702 in the neighborhood point extraction unit 2221.

  On the other hand, if the feature amount calculation unit 222 determines that the feature amount calculated by the feature amount extraction unit 2222 is valid (YES in S704), the feature amount calculation unit 222 outputs the calculated feature amount to the memory 225, and ends the process. The fifth embodiment has been described above.

  In the present embodiment, the feature amount calculation unit 222 has been described as an example in which the peripheral region setting process, the neighboring feature point extraction process, and the feature amount calculation process are performed in series. A series of processes may be performed based on different mask sizes or reference block numbers, a process having the largest number of neighboring feature points may be selected, and a feature amount may be calculated based on data by the selected process. This will be described below as Example 6.

Example 6
FIG. 19 is a schematic diagram illustrating a configuration of the feature amount calculation unit 222 according to the sixth embodiment. In the sixth embodiment, the pixel block setting unit 2220 is a first pixel block setting unit 2220a and a second pixel block setting unit 2220b, and the neighboring point extracting unit 2221 is a first neighboring point extracting unit 2221a and a second neighboring point extracting unit 2221b. Further, a distribution unit 2223 and a selection unit 2224 are provided. The distribution unit 2223 receives the feature point data stored in the memory 225, and allocates the received data to the first pixel block setting unit 2220a and the second pixel block setting unit 2220b. Each of the first pixel block setting unit 2220a (and the second pixel block setting unit 2220b) sets a pixel block composed of one or a plurality of pixels of the image, and sets the mask size or the number of reference blocks of the set pixel block. , One of the extracted feature points as a target feature point, a neighboring pixel block centered on the pixel block including the target feature point is set as a “peripheral region”, and the set pixel block, mask size or reference block The data including the number and the surrounding area is output to the first neighboring point extracting unit 2221a (and the second neighboring point extracting unit 2221b). The first neighboring point extraction unit 2221a (and the second neighboring point extraction unit 2221b) receives the data output from the first pixel block setting unit 2220a (and the second pixel block setting unit 2220b), respectively, and sets the received data Based on this, the feature points in the vicinity located in the surrounding area are extracted, the number of the feature points in the extracted neighborhood is counted, whether the counted number is equal to or less than the second threshold value, and the determination result is selected. Output to the unit 2224. The selection unit 2224 receives the determination result output from the first neighboring point extraction unit 2221a (and the second neighboring point extraction unit 2221b), and determines that the counted number exceeds the second threshold value among the received determination results. (If both exceed the second threshold value, the one with the higher score) is selected, and the data on the neighboring feature points is transferred from the neighboring point extraction unit that derived the selected determination result to the feature amount extraction unit 2222. Output. The feature amount extraction unit 2222 receives the data output from the neighborhood point extraction unit, calculates the feature amount based on the received data, and outputs the calculated feature amount to the memory 225. The sixth embodiment has been described above.

  Further, in the present embodiment, an example in which the number of reference blocks is changed without counting the center of gravity in units of pixel block lines has been described, but not limited thereto, the center of gravity is counted in units of pixel block lines, The number of reference blocks may be changed according to the result. This will be described below as Example 7.

Example 7
In the seventh embodiment, since the configuration of the centroid calculating unit 2214 for counting the centroid in units of pixel block lines is the same as that in FIG. 34, the same reference numerals are given and the description thereof is omitted.

  FIG. 38 is a flowchart illustrating a processing procedure of the feature amount calculation unit 222 according to the seventh embodiment. The feature amount calculation unit 222 first sets the number of blocks to be referred to in units of pixel blocks stored in the centroid count buffer 2214f (S901).

  The feature amount calculation unit 222 reads out the centroid count result corresponding to the pixel block line that performs the processing in which the pixel block including the target feature point and the block to be referenced exist, from the centroid count buffer 2214f (S902), and the read pixel The centroid count result of the block line, that is, the reference block line is added (S903).

  The feature amount calculation unit 222 determines whether or not the addition result is smaller than the third threshold (S904). Here, when four feature points in the vicinity are extracted, the feature quantity cannot be accurately calculated unless there are four or more feature points around the target feature point. Therefore, it is necessary to change the number of reference blocks.

  If the feature amount calculation unit 222 determines that the addition result is smaller than the third threshold value (YES in S904), the feature point calculation unit 222 changes to a larger reference block number because there is no neighboring feature point necessary for calculating the feature point. In step S905, the process returns to step S902 to read data from the centroid count buffer 2214f.

  On the other hand, when it is determined that the addition result is equal to or greater than the third threshold (NO in S904), the feature amount calculation unit 222 extracts nearby feature points (S906) and outputs the feature points to the feature amount extraction unit 2222. The feature amount extraction unit 2222 calculates a feature amount based on the received data (S907). The feature amount calculation unit 222 ends the process.

  FIG. 39 is a schematic diagram for explaining threshold determination and changing the number of reference blocks. In FIG. 39, the reference block number setting is set to 9 (3 × 3 mask). In this case, since the count number BCCNT3 = 1 of the pixel block line including the feature point of interest and the count numbers of the pixel block lines of the reference block are BCCNT2 = 1 and BCCNT4 = 2, respectively, the addition result is 4. The third threshold value (TH_SPREAD) is 4 neighboring feature points + one feature point of interest = 5. Accordingly, the feature amount calculation unit 222 determines that the addition result is smaller than the third threshold (refer to YES in S904), and sets the reference block number 9 (3 × 3 mask) initially set to 25 (5 × 5 mask). Thereby, four neighborhood feature points can be secured.

  Note that the document image described above is divided, a lower limit value of a connected area is set in each divided area, and a process including a pixel block including a point of interest and its surrounding blocks is set in each divided area. The centroid count result corresponding to the pixel block line to be performed may be read from the centroid count buffer 2214f and added, and when the addition result is equal to or smaller than the second threshold, the number of reference blocks may be changed.

  When the centroid count result corresponding to the pixel block including the target point and the pixel block line that performs processing including the surrounding blocks is read from the centroid count buffer 2214f and added, and the addition result is smaller than the third threshold value Alternatively, the lower limit value of the number of pixels for extracting a connected area for obtaining a feature point or the lower limit value of the number of pixels for extracting a connected area for obtaining a feature point and the number of reference blocks may be changed.

  Furthermore, a plurality of lower limit values of the number of pixels for extracting a connected region for obtaining a feature point may be set, and the reference block number may be changed when the counted most feature points are smaller than the third threshold value. The seventh embodiment has been described above.

  Next, the feature amount calculation process executed by the feature amount calculation unit 222 in step S406 and the like will be described. FIG. 20 is an explanatory diagram showing the relationship between the feature point of interest and neighboring feature points, FIGS. 21 and 22 are explanatory diagrams showing examples of invariant calculation using the feature point of interest, and FIG. 23 is an explanatory diagram showing the structure of the hash table. It is.

  Based on the data received from the memory 225 by the feature amount extraction unit 2222, the feature amount calculation unit 222, for example, sets four feature points located in the peripheral area S 1, for example, as the attention feature. Extraction is performed in order of increasing distance from the point P1 (the feature point of interest P2 is a nearby feature point). Also, with respect to the feature point P2, four feature points located in the peripheral region S2 are extracted in order of increasing distance from the feature point P2 as described above (the feature point P1 is defined as a nearby feature point). To do).

  The feature amount calculation unit 222 selects three feature points from the four feature points extracted by the feature amount extraction unit 2222 and calculates an invariant by a method described later. Note that the number of feature points to be selected is not limited to three, and four or five feature points may be selected.

  Three neighboring feature points are selected from the four neighboring feature points located in the peripheral region of the target feature point P1, and each of the three invariants is set to H1j (j = 1, 2, 3). For the invariant H1j, the invariants H11, H12, and H13 are obtained using the formula of H1j = A1j / B1j (see FIG. 21). A1j and B1j indicate distances between feature points, and the distances between feature points are calculated based on the coordinate values of each peripheral feature point. Thereby, even when the image is rotated, moved, or tilted, the invariant H11 and the like are constant, so that the accuracy of similarity determination of images using the invariant can be stabilized.

  Also, three neighboring feature points are selected from the four neighboring feature points located in the peripheral region of the target feature point P2, and each of the three invariants is represented as H2j (j = 1, 2, 3). To do. For the invariant H2j, the invariants H21, H22, and H23 are obtained using the formula H2j = A2j / B2j (see FIG. 22). A2j and B2j indicate the distances between the feature points, and the distances between the feature points are calculated based on the coordinate values of the peripheral feature points in the same manner as described above. Thereby, even when the image is rotated, moved, or tilted, the invariant H11 and the like are constant, so that the accuracy of similarity determination of images using the invariant can be stabilized.

In addition, the feature amount calculation unit 222 calculates a hash value (feature amount) Hij based on the invariant by the feature amount extraction unit 2222. The hash value Hij is Hij = (Hi1 × 10 ² + Hi
2 × 10 ¹ + Hi3 × 10 ⁰ ) / E. However, i is a natural number and represents the number of feature points, and E is a constant determined by how much remainder is set. For example, when E = 10, the remainder is 1 to 9, which is the hash value range to be calculated.

FIG. 26 and FIG. 27 are explanatory diagrams showing an example of calculating invariants based on the feature point of interest. As a method for calculating the invariant based on the feature point of interest, for example, as shown in FIG. 26, four combinations are selected from the four feature points P1, P2, P4, and P5 around the feature point of interest P3. The variable H3j (j = 1, 2, 3, 4) may be calculated by H3j = A3j / B3j in the same manner as described above. Similarly, when the target feature point is set to P4, four combinations are selected from the four peripheral feature points P2, P3, P5, and P6 of the target feature point P4 (see FIG. 27), and the invariant H4j ( j = 1, 2, 3, 4) may be calculated by H4j = A4j / B4j. In this case, the hash value Hi is calculated as Hi = (Hi1 × 10 ³ + Hi2 × 10 ² + Hi3 × 10 ¹ + Hi4 × 10 ⁰ ) / E.
Note that the hash value as the feature amount is an example, and other hash functions can be used. In the above example, the hash value is calculated based on three neighboring feature points. However, the present invention is not limited to this, and the hash value may be calculated based on four or five neighboring feature points. Good.

  The calculated hash value is stored in a hash table in association with an index (see FIG. 23A). In the hash table, image data of the other document to be collated is registered in advance, and a hash value is registered in association with an index representing the document. The hash table is stored in the memory 225. When the hash values are the same (for example, H11 and H22), the corresponding two entries may be combined into one (see FIG. 23B). The image data of the other document to be collated is received in advance via the image input device 1 and registered in the hash table.

  Next, the similarity determination process executed by the voting processing unit 223 and the similarity determination processing unit 224 will be described.

  The voting processing unit 223 receives the feature amount (hash value) output from the feature amount calculation unit 222, searches the hash table stored in the memory 225 based on the received feature amount, and corresponds to the feature amount. The index document is voted, and the number of votes obtained by accumulating the votes is output to the similarity determination processing unit 224.

  The similarity determination processing unit 224 receives the number of votes output from the voting processing unit 223, determines whether or not the received number of votes is equal to or greater than a set threshold, and determines that the number of votes is equal to or greater than the threshold. A determination result indicating that the image to be determined this time is similar to the image of the other document registered in the hash table is output to the memory 225. In addition, when the number of votes is extremely larger than the threshold value, the similarity determination processing unit 224 transmits a determination result indicating that the image to be determined this time matches the image of the other document registered in the hash table to the memory 225. Output. On the other hand, when the similarity determination processing unit 224 determines that the number of votes is less than the threshold, the determination result that the image to be determined this time is not similar to the image of the other document registered in the hash table is output. To do. The control unit executes predetermined processing such as image output prohibition, copy prohibition, and storage in a predetermined folder according to the determination result.

  Note that the similarity determination processing of the similarity determination processing unit 224 is an example. For example, the number of votes obtained may be divided and normalized by the maximum number of votes for each document, and the similarity determination may be performed based on the result.

  In the above-described embodiment, the feature point calculation unit 221 adjusts the number of centroids (feature points) to be calculated by performing threshold determination in the centroid calculation unit 2214, and further, the mask size of the pixel block or the number of reference blocks In the example described above, the number of feature points to be referred to is adjusted (see FIGS. 6, 14, etc.), but not limited to this, the adjustment of the number of feature points is performed by the feature point calculation unit 221. The center-of-gravity calculation unit 2214 may simply adjust the number of centers of gravity (feature points) to be calculated by performing threshold determination.

  In the embodiment described above, the registered image and the hash table are stored in the memory 225 in advance. However, the present invention is not limited to this, and the server device connected to the image forming apparatus through a communication line (network). The registered image may be stored in the storage unit, and the hash table may be distributed and stored in the memory 225.

  Each unit (each block) constituting the image matching processing unit 22 and the control unit 226 provided in the image forming apparatus is realized by software using a processor such as a CPU.

That is, the image forming apparatus includes a central processing unit (CPU) that executes a command of a control program that realizes each function, and a read only memory (ROM) that stores the control program.
A random access memory (RAM) for developing the control program, a storage device (recording medium) such as a memory for storing the control program and various data, and the like (not shown) are provided.

  The object of the present invention is achieved by executing the steps (execution format program, intermediate code program, source program) included in the program code of the control program of the multifunction machine, which is software that implements the above-described functions. In addition, a computer-readable recording medium in which the control program is recorded is inserted into the image forming apparatus, and the computer (or CPU or MPU) reads and executes the program code recorded in the recording medium. Is done.

  As a recording medium, for example, a tape system such as a magnetic tape or a cassette tape, a disk system including a magnetic disk such as a floppy (registered trademark) disk / hard disk or an optical disk such as a CD-ROM / MO / MD / DVD / CD-R, This includes card systems such as IC cards (including memory cards) / optical cards, and semiconductor memory systems such as mask ROM / EPROM / EEPROM / flash ROM.

Further, the image forming apparatus may be connected to a communication network, and the program may be supplied via the communication network. The communication network is not particularly limited. For example, the Internet, intranet, extranet, LAN, ISDN, VAN, CATV communication network, virtual private network, telephone line network, mobile communication network, satellite communication. This applies to the net. Also, the transmission medium constituting the communication network is not particularly limited. For example, IEEE1394, USB, power line carrier, cable TV line, telephone line, ADSL line, etc., infrared such as IrDA or remote control, Bluetooth (registration) Trademark), 802.11 radio, HDR, mobile phone network, satellite line, terrestrial digital network, and the like are applicable. The present invention can also be realized in the form of a computer data signal embedded in a carrier wave, in which the program code is embodied by electronic transmission.

  In addition, each block of the image forming apparatus is not limited to being realized using software, and may be configured by hardware logic. Hardware that performs a part of processing and control of the hardware Or a calculation unit that executes software for performing the remaining processing.

  The computer system according to the present invention performs various processes such as an image input device such as a flatbed scanner, a film scanner, and a digital camera, and a similarity calculation process and a similarity determination process by loading a predetermined program. You may comprise by image formation prosecution, such as a computer, image display apparatuses, such as a CRT display and a liquid crystal display which display the processing result of a computer, and a printer which outputs the processing result of a computer on paper. Furthermore, a network card, a modem, or the like as communication means for connecting to a server or the like via a network may be provided.

1 is a schematic diagram illustrating an overall configuration of an image forming apparatus including an image processing apparatus according to the present invention. It is a schematic diagram which shows the whole structure of an image collation process part. It is a schematic diagram which shows the structure of a feature point calculation part. It is a schematic diagram which shows the structure of a gravity center calculation part. It is a flowchart which shows the procedure of the gravity center calculation process of a gravity center calculation part. It is a flowchart which shows the procedure of the threshold value determination of a gravity center calculation part. It is a flowchart which shows the procedure of the addition process of a gravity center calculation part. It is explanatory drawing which shows an example of threshold value determination. FIG. 5 is an explanatory diagram illustrating an example of a document with few feature points. 3 is a schematic diagram illustrating a configuration of a centroid calculating unit according to Embodiment 1. FIG. It is a schematic diagram which shows the structure of a feature-value calculation part. It is explanatory drawing which shows the peripheral region of an attention feature point. It is explanatory drawing which shows the relationship between the mask size of a peripheral region, and the nearby feature point located in this peripheral region. It is a flowchart which shows the procedure of the process of a feature-value calculation part. 12 is a flowchart illustrating a processing procedure of a feature amount calculation unit according to the third embodiment. It is explanatory drawing which shows the relationship between the pixel block number of a peripheral region, and the nearby feature point located in this peripheral region. 12 is a flowchart illustrating a processing procedure of a feature amount calculation unit according to the third embodiment. 16 is a flowchart illustrating a processing procedure of a feature amount calculation unit according to the fifth embodiment. FIG. 10 is a schematic diagram illustrating a configuration of a feature amount calculation unit according to a sixth embodiment. It is explanatory drawing which shows the relationship between an attention feature point and the nearby feature point. It is explanatory drawing which shows the example of calculation of the invariant by an attention feature point. It is explanatory drawing which shows the example of calculation of the invariant by an attention feature point. It is explanatory drawing which shows the structure of a hash table. It is explanatory drawing which shows the gravity center of a connection component. It is explanatory drawing which shows the example which makes a gravity center a feature point. It is explanatory drawing which shows the example of calculation of the invariant by an attention feature point. It is explanatory drawing which shows the example of calculation of the invariant by an attention feature point. It is explanatory drawing which shows an example of the original document image input. It is explanatory drawing which shows an example which equally divided the original image into 4 parts. FIG. 10 is a schematic diagram illustrating a configuration of a gravity center calculation unit according to a second embodiment. It is explanatory drawing which shows the structure of a threshold value table. It is explanatory drawing which shows the structure of a threshold value table. It is explanatory drawing which shows an example which divided | segmented the original image. It is a schematic diagram which shows the structure of the gravity center calculation part for counting a gravity center in a pixel block line unit. It is explanatory drawing which shows the example of the count by a gravity center count buffer. 14 is a flowchart illustrating a processing procedure of a feature amount calculation unit according to the fourth embodiment. It is a schematic diagram explaining threshold value determination and mask size change. 18 is a flowchart illustrating a processing procedure of a feature amount calculation unit according to the seventh embodiment. It is a schematic diagram explaining threshold value determination and the change of the number of reference blocks.

Explanation of symbols

2 image processing apparatus 22 image collation processing unit 221 feature point calculation unit 2213 binarization processing unit 2214 centroid calculation unit 2214a labeling processing unit 2214b connected component threshold processing unit 2214c centroid calculation processing unit 2214d centroid storage buffer 2214e control unit 2214f centroid count buffer 222 feature amount calculation unit 2220 pixel block setting unit 2221 neighborhood point extraction unit 2222 feature amount extraction unit 224 similarity determination processing unit 225 memory 226 control unit

Claims (31)

A plurality of pixel regions adjacent to pixels determined to have the same pixel value from a binary image are identified, and feature points of the pixel region are extracted based on the coordinate values of the pixels of the identified pixel region and extracted. In an image processing method for calculating a feature amount indicating a feature of an image based on a feature point, and performing similarity determination between images based on the calculated feature amount,
Count the number of pixels in the specified pixel area,
Determining whether the number of pixels in the identified pixel region is greater than or equal to a first threshold;
When it is determined that the number of pixels in the identified pixel area is equal to or greater than the first threshold, the feature points of the pixel area are calculated and the number of feature points is counted,
Determine whether the number of feature points counted is less than or equal to a second threshold;
When it is determined that the number of counted feature points exceeds the second threshold, the feature amount is calculated based on the feature points extracted from the pixel region,
An image processing method comprising: changing the first threshold value when it is determined that the number of counted feature points is equal to or less than a second threshold value. A plurality of pixel regions adjacent to pixels determined to have the same pixel value from a binary image are identified, and feature points of the pixel region are extracted based on the coordinate values of the pixels of the identified pixel region and extracted. In an image processing method for calculating a feature amount indicating a feature of an image based on a feature point, and performing similarity determination between images based on the calculated feature amount,
Count the number of pixels in the specified pixel area,
When determining whether or not the number of pixels in the identified pixel region is equal to or greater than the first threshold, a plurality of different first thresholds are set,
Determine whether each is greater than or equal to the first threshold,
If it is determined that it is equal to or greater than the first threshold, the feature points of the pixel region are calculated, and the number of feature points is counted for each of a plurality of different first thresholds,
An image processing method, wherein a feature amount is calculated based on a feature point extracted from a pixel region having the largest number of counted feature points. A plurality of pixel regions adjacent to pixels determined to have the same pixel value from a binary image are identified, and feature points of the pixel region are extracted based on the coordinate values of the pixels of the identified pixel region and extracted. In an image processing method for calculating a feature amount indicating a feature of an image based on a feature point, and performing similarity determination between images based on the calculated feature amount,
Divide the original image area into multiple areas,
Set a first threshold for each divided region,
In each divided area, count the number of pixels in the specified pixel area,
Determining whether the number of pixels in the pixel area identified in the divided area is equal to or greater than a first threshold in the divided area;
When it is determined that the number of pixels in the pixel area specified in the divided area is equal to or greater than a first threshold in the divided area, the feature points of the pixel area are calculated and the number of feature points is counted,
Determine whether the number of feature points counted is less than or equal to a second threshold;
When it is determined that the number of counted feature points exceeds the second threshold, the feature amount is calculated based on the feature points extracted from the pixel region,
An image processing method comprising: changing the first threshold value when it is determined that the number of counted feature points is equal to or less than a second threshold value. A plurality of pixel regions adjacent to pixels determined to have the same pixel value from a binary image are identified, and feature points of the pixel region are extracted based on the coordinate values of the pixels of the identified pixel region and extracted. In an image processing method for calculating a feature amount indicating a feature of an image based on a feature point, and performing similarity determination between images based on the calculated feature amount,
Determine nearby feature points located within the surrounding area of the extracted feature points,
Count the number of feature points in the neighborhood
Determine whether the number of feature points counted is less than or equal to the second threshold;
When it is determined that the counted number of feature points exceeds the second threshold, the feature amount is calculated based on the neighboring feature points;
An image processing method comprising: changing the range of the peripheral area when it is determined that the counted number is equal to or less than a second threshold value. A plurality of pixel regions adjacent to pixels determined to have the same pixel value from a binary image are identified, and feature points of the pixel region are extracted based on the coordinate values of the pixels of the identified pixel region and extracted. In an image processing method for calculating a feature amount indicating a feature of an image based on a feature point, and performing similarity determination between images based on the calculated feature amount,
When extracting feature points, count the number of feature points within a predetermined range,
Extract the feature points included in the area to be processed from the counted feature points,
Determine whether the number of extracted feature points is smaller than the third threshold;
If it is determined that the number of extracted feature points is greater than or equal to the third threshold, the feature amount is calculated based on nearby feature points located in the peripheral region of the extracted feature points;
An image processing method characterized by changing the range of the peripheral area when it is determined that the extracted score is smaller than a third threshold. A plurality of pixel regions adjacent to pixels determined to have the same pixel value from a binary image are identified, and feature points of the pixel region are extracted based on the coordinate values of the pixels of the identified pixel region and extracted. In an image processing method for calculating a feature amount indicating a feature of an image based on a feature point, and performing similarity determination between images based on the calculated feature amount,
Count the number of pixels in the specified pixel area,
Determining whether the number of pixels in the identified pixel region is greater than or equal to a first threshold;
If it is determined that the number of pixels in the identified pixel area is equal to or greater than the first threshold, the feature points of the pixel area are extracted,
Determine nearby feature points located within the surrounding area of the extracted feature points,
Count the number of feature points in the neighborhood
Determine whether the number of feature points counted is less than or equal to the second threshold;
When it is determined that the counted number of feature points exceeds the second threshold, the feature amount is calculated based on the neighboring feature points;
An image processing method characterized by changing the first threshold value or the first threshold value and the range of the surrounding area when it is determined that the number of counted feature points is equal to or less than the second threshold value. . A plurality of pixel regions adjacent to pixels determined to have the same pixel value from a binary image are identified, and feature points of the pixel region are extracted based on the coordinate values of the pixels of the identified pixel region and extracted. In an image processing method for calculating a feature amount indicating a feature of an image based on a feature point, and performing similarity determination between images based on the calculated feature amount,
Count the number of pixels in the specified pixel area,
When determining whether or not the number of pixels in the identified pixel region is equal to or greater than the first threshold, a plurality of different first thresholds are set,
Determine whether each is greater than or equal to the first threshold,
If it is determined that it is equal to or greater than the first threshold, the feature points of the pixel region are calculated, and the number of feature points is counted for each of a plurality of different first thresholds,
Extract the feature points of the pixel area where the number of counted feature points is the largest,
Determine nearby feature points located within the surrounding area of the extracted feature points,
Count the number of feature points in the neighborhood
Determine whether the number of feature points counted is less than or equal to the second threshold;
When it is determined that the counted number of feature points exceeds the second threshold, the feature amount is calculated based on the neighboring feature points;
An image processing method, comprising: changing a range of the surrounding area when it is determined that the number of counted feature points is equal to or less than a second threshold value.   The peripheral area is composed of pixel blocks including one or a plurality of pixels of a binary image, and the range of the peripheral area is changed by changing the size or the number of blocks of the pixel block. The image processing method according to any one of claims 4 to 7. A plurality of pixel regions adjacent to pixels determined to have the same pixel value from a binary image are identified, and feature points of the pixel region are extracted based on the coordinate values of the pixels of the identified pixel region and extracted. In an image processing apparatus that calculates a feature amount indicating an image feature based on a feature point and performs similarity determination between images based on the calculated feature amount.
Means for counting the number of pixels in the identified pixel region;
Means for determining whether or not the number of pixels in the identified pixel region is equal to or greater than a first threshold;
Means for calculating the feature points of the pixel region and counting the number of feature points when it is determined that the number of pixels in the identified pixel region is equal to or greater than the first threshold;
Means for determining whether or not the counted number of feature points is equal to or less than a second threshold;
When it is determined that the counted number of feature points exceeds the second threshold value, the feature amount is calculated based on the feature points extracted from the pixel area, and the counted number of feature points is equal to or less than the second threshold value. If it is determined that the first threshold value is changed, the image processing apparatus is characterized in that the first threshold value is changed. A plurality of pixel regions adjacent to pixels determined to have the same pixel value from a binary image are identified, and feature points of the pixel region are extracted based on the coordinate values of the pixels of the identified pixel region and extracted. In an image processing apparatus that calculates a feature amount indicating an image feature based on a feature point and performs similarity determination between images based on the calculated feature amount.
Means for counting the number of pixels in the identified pixel region;
Determining means for determining whether or not the number of specified pixel regions is equal to or greater than a first threshold, and setting a plurality of different first thresholds and determining whether or not each is greater than or equal to the first threshold; ,
Means for calculating the feature points of the pixel region when it is determined that the threshold value is greater than or equal to the first threshold, and counting the number of feature points for each of a plurality of different first thresholds;
An image processing apparatus characterized in that a feature amount is calculated based on a feature point extracted from a pixel region where the number of counted feature points is the largest. A plurality of pixel regions adjacent to pixels determined to have the same pixel value from a binary image are identified, and feature points of the pixel region are extracted based on the coordinate values of the pixels of the identified pixel region and extracted. In an image processing apparatus that calculates a feature amount indicating an image feature based on a feature point and performs similarity determination between images based on the calculated feature amount.
Means for dividing a document image area into a plurality of areas;
Means for setting a first threshold for each of the divided areas;
Means for counting the number of pixels in the specified pixel area in each of the divided areas;
Means for determining whether or not the number of pixels in the pixel area specified in the divided area is equal to or greater than a first threshold in the divided area;
Means for calculating the feature points of the pixel region and counting the number of feature points when it is determined that the number of pixels in the pixel region specified in the divided region is equal to or greater than a first threshold value in the divided region; When,
Means for determining whether or not the counted number of feature points is equal to or less than a second threshold;
When it is determined that the counted number of feature points exceeds the second threshold value, the feature amount is calculated based on the feature points extracted from the pixel area, and the counted number of feature points is equal to or less than the second threshold value. If it is determined that the first threshold value is changed, the image processing apparatus is characterized in that the first threshold value is changed. A plurality of pixel regions adjacent to pixels determined to have the same pixel value from a binary image are identified, and feature points of the pixel region are extracted based on the coordinate values of the pixels of the identified pixel region and extracted. In an image processing apparatus that calculates a feature amount indicating an image feature based on a feature point and performs similarity determination between images based on the calculated feature amount.
Means for determining neighboring feature points located within the peripheral region of the extracted feature points;
Means for counting the number of feature points in the determined neighborhood;
Means for determining whether or not the number of counted feature points is equal to or less than a second threshold value,
When it is determined that the number of feature points counted exceeds the second threshold, the feature amount is calculated based on the neighboring feature points, and it is determined that the score of the counted feature points is equal to or less than the second threshold. In some cases, the range of the peripheral region is changed. A plurality of pixel regions adjacent to pixels determined to have the same pixel value from a binary image are identified, and feature points of the pixel region are extracted based on the coordinate values of the pixels of the identified pixel region and extracted. In an image processing apparatus that calculates a feature amount indicating an image feature based on a feature point and performs similarity determination between images based on the calculated feature amount.
Means for counting the number of feature points within a predetermined range when extracting feature points;
Means for extracting feature points included in the region to be processed from the counted feature points;
Means for determining whether or not the number of extracted feature points is smaller than a third threshold;
If it is determined that the number of extracted feature points is greater than or equal to the third threshold, the feature amount is calculated based on neighboring feature points located in the peripheral region of the extracted feature points, and the extracted number of points is An image processing apparatus, wherein when it is determined that the threshold value is smaller than 3, the range of the peripheral area is changed. A plurality of pixel regions adjacent to pixels determined to have the same pixel value from a binary image are identified, and feature points of the pixel region are extracted based on the coordinate values of the pixels of the identified pixel region and extracted. In an image processing apparatus that calculates a feature amount indicating an image feature based on a feature point and performs similarity determination between images based on the calculated feature amount.
Means for counting the number of pixels in the identified pixel region;
Means for determining whether or not the number of pixels in the identified pixel region is equal to or greater than a first threshold;
Means for extracting a feature point of the pixel area when it is determined that the number of pixels in the identified pixel area is equal to or greater than a first threshold;
Means for determining neighboring feature points located within the peripheral region of the extracted feature points;
Means for counting the number of feature points in the determined neighborhood;
Means for determining whether or not the number of counted feature points is equal to or less than a second threshold value,
When it is determined that the score of the counted feature points exceeds the second threshold, the feature amount is calculated based on the neighboring feature points, and the score of the counted feature points is equal to or less than the second threshold. The image processing apparatus according to claim 1, wherein the first threshold value or the first threshold value and the range of the peripheral region are changed when the determination is made. A plurality of pixel regions adjacent to pixels determined to have the same pixel value from a binary image are identified, and feature points of the pixel region are extracted based on the coordinate values of the pixels of the identified pixel region and extracted. In an image processing apparatus that calculates a feature amount indicating an image feature based on a feature point and performs similarity determination between images based on the calculated feature amount.
Means for counting the number of pixels in the identified pixel region;
When determining whether or not the number of pixels in the specified pixel area is equal to or greater than the first threshold, a plurality of different first thresholds are set, and determination is performed to determine whether or not each is greater than or equal to the first threshold Means,
Means for calculating the feature points of the pixel region when it is determined to be greater than or equal to the first threshold, and counting the number of feature points to a plurality of different first thresholds;
Means for extracting feature points of the pixel region where the number of counted feature points is the largest;
Means for determining neighboring feature points located within the peripheral region of the extracted feature points;
Means for counting the number of feature points in the determined neighborhood;
Means for determining whether or not the number of counted feature points is equal to or less than a second threshold value,
When it is determined that the number of feature points counted exceeds the second threshold, the feature amount is calculated based on the neighboring feature points, and it is determined that the score of the counted feature points is equal to or less than the second threshold. In some cases, the range of the peripheral region is changed.   The peripheral area is composed of pixel blocks composed of one or a plurality of pixels of a binary image, and the range of the peripheral area is changed by changing the size or the number of blocks of the pixel block. The image processing apparatus according to claim 12, wherein the image processing apparatus is an image processing apparatus. An image processing apparatus according to any one of claims 9 to 16,
An image forming apparatus comprising: an image forming unit that forms an image processed by the image processing apparatus. Let a computer specify a plurality of adjacent pixel regions that are judged to have the same pixel value from a binary image, and extract feature points of the pixel region based on the coordinate values of the pixels of the specified pixel region In a computer program for calculating a feature amount indicating the feature of an image based on the extracted feature point, and determining similarity between images based on the calculated feature amount,
Causing the computer to count the number of pixels in the identified pixel area;
Causing the computer to determine whether the number of pixels in the identified pixel region is greater than or equal to a first threshold;
When the computer determines that the number of pixels in the specified pixel region is equal to or greater than the first threshold, calculating the feature points of the pixel region and counting the number of feature points;
Causing the computer to determine whether the number of feature points counted is less than or equal to a second threshold;
Causing the computer to calculate the feature amount based on the feature points extracted from the pixel area when it is determined that the number of counted feature points exceeds a second threshold;
A computer program for causing a computer to execute a step of changing the first threshold value when it is determined that the number of counted feature points is equal to or less than a second threshold value. Let a computer specify a plurality of adjacent pixel regions that are judged to have the same pixel value from a binary image, and extract feature points of the pixel region based on the coordinate values of the pixels of the specified pixel region In a computer program for calculating a feature amount indicating the feature of an image based on the extracted feature point, and determining similarity between images based on the calculated feature amount,
Causing the computer to count the number of pixels in the identified pixel area;
When determining whether or not the number of pixels in the specified pixel region is greater than or equal to the first threshold, the computer sets a plurality of different first thresholds and determines whether or not each is greater than or equal to the first threshold. A step to make a decision;
A step of causing the computer to calculate the feature points of the pixel region and to count the number of feature points for each of a plurality of different first threshold values when it is determined that the value is equal to or greater than the first threshold value;
A computer program for causing a computer to calculate a feature amount based on a feature point extracted from a pixel region having the largest number of counted feature points. Let a computer specify a plurality of adjacent pixel regions that are judged to have the same pixel value from a binary image, and extract feature points of the pixel region based on the coordinate values of the pixels of the specified pixel region In a computer program for calculating a feature amount indicating the feature of an image based on the extracted feature point, and determining similarity between images based on the calculated feature amount,
Causing the computer to divide the document image area into a plurality of areas;
Causing the computer to set a first threshold for each divided region;
Causing the computer to count the number of pixels in the specified pixel area in each divided area;
Causing the computer to determine whether or not the number of pixels in the identified pixel area is equal to or greater than a first threshold in the divided area;
When the computer determines that the number of pixels in the specified pixel area in the divided areas is equal to or greater than the set first threshold, the computer calculates feature points of the pixel area and counts the number of feature points Steps,
Causing the computer to determine whether the number of feature points counted is less than or equal to a second threshold;
Causing the computer to calculate the feature amount based on the feature points extracted from the pixel area when it is determined that the number of counted feature points exceeds a second threshold;
A computer program for causing a computer to execute a step of changing the first threshold value when it is determined that the number of counted feature points is equal to or less than a second threshold value. Let a computer specify a plurality of adjacent pixel regions that are judged to have the same pixel value from a binary image, and extract feature points of the pixel region based on the coordinate values of the pixels of the specified pixel region In a computer program for calculating a feature amount indicating the feature of an image based on the extracted feature point, and determining similarity between images based on the calculated feature amount,
Causing a computer to determine nearby feature points located within a peripheral region of the extracted feature points;
Causing the computer to count the number of nearby feature points determined;
Causing the computer to determine whether the counted number of feature points is equal to or less than a second threshold;
If the computer determines that the counted number of feature points exceeds a second threshold, calculating the feature amount based on the nearby feature points;
A computer program for causing a computer to execute a step of changing a range of the peripheral area when it is determined that the number of counted feature points is equal to or less than a second threshold value. Let a computer specify a plurality of adjacent pixel regions that are judged to have the same pixel value from a binary image, and extract feature points of the pixel region based on the coordinate values of the pixels of the specified pixel region In a computer program for calculating a feature amount indicating the feature of an image based on the extracted feature point, and determining similarity between images based on the calculated feature amount,
Causing the computer to count feature points within a predetermined range when extracting feature points; and
Allowing the computer to determine whether the number of extracted feature points is less than a third threshold;
When the computer determines that the number of extracted feature points is equal to or greater than a third threshold, calculating the feature amount based on neighboring feature points located in a peripheral region of the extracted feature points; ,
A computer program for causing a computer to execute a step of changing a range of the peripheral area when it is determined that the extracted score is smaller than a third threshold value. Let a computer specify a plurality of adjacent pixel regions that are judged to have the same pixel value from a binary image, and extract feature points of the pixel region based on the coordinate values of the pixels of the specified pixel region In a computer program for calculating a feature amount indicating the feature of an image based on the extracted feature point, and determining similarity between images based on the calculated feature amount,
Causing the computer to count the number of pixels in the identified pixel area;
Causing the computer to determine whether the number of pixels in the identified pixel region is greater than or equal to a first threshold;
If the computer determines that the number of pixels in the specified pixel area is equal to or greater than the first threshold, extracting a feature point of the pixel area;
Causing a computer to determine nearby feature points located within a peripheral region of the extracted feature points;
Causing the computer to count the number of nearby feature points determined;
Causing the computer to determine whether the counted number of feature points is equal to or less than a second threshold;
Causing the computer to calculate the feature amount based on the neighboring feature points when it is determined that the score of the counted feature points exceeds the second threshold;
Causing the computer to change the first threshold value or the first threshold value and the range of the surrounding area when it is determined that the number of feature points counted is less than or equal to the second threshold value. A computer program characterized by the above. Let a computer specify a plurality of adjacent pixel regions that are judged to have the same pixel value from a binary image, and extract feature points of the pixel region based on the coordinate values of the pixels of the specified pixel region In a computer program for calculating a feature amount indicating the feature of an image based on the extracted feature point, and determining similarity between images based on the calculated feature amount,
Causing the computer to count the number of pixels in the identified pixel area;
When determining whether or not the number of pixels in the specified pixel region is greater than or equal to the first threshold, the computer sets a plurality of different first thresholds and determines whether or not each is greater than or equal to the first threshold. A step for judging,
A step of causing the computer to calculate the feature points of the pixel region and to count the number of feature points for each of a plurality of different first threshold values when it is determined that the value is equal to or greater than the first threshold value;
Causing the computer to extract the feature points of the pixel region having the largest number of feature points counted;
Causing a computer to determine nearby feature points located within a peripheral region of the extracted feature points;
Letting the computer score points of the determined nearby feature points;
Causing the computer to determine whether the counted number of feature points is equal to or less than a second threshold;
If the computer determines that the counted number of feature points exceeds a second threshold, calculating the feature amount based on the nearby feature points;
A computer program for causing a computer to execute a step of changing a range of the peripheral area when it is determined that the number of counted feature points is equal to or less than a second threshold value. Let a computer specify a plurality of adjacent pixel regions that are judged to have the same pixel value from a binary image, and extract feature points of the pixel region based on the coordinate values of the pixels of the specified pixel region , A computer-readable recording medium in which a computer program for calculating a feature amount indicating an image feature based on the extracted feature point and determining similarity between images based on the calculated feature amount is recorded In
Causing the computer to count the number of pixels in the identified pixel area;
Causing the computer to determine whether the number of pixels in the identified pixel region is greater than or equal to a first threshold;
When the computer determines that the number of pixels in the specified pixel region is equal to or greater than the first threshold, calculating the feature points of the pixel region and counting the number of feature points;
Causing the computer to determine whether the number of feature points counted is less than or equal to a second threshold;
Causing the computer to calculate the feature amount based on the feature points extracted from the pixel area when it is determined that the number of counted feature points exceeds a second threshold;
And a computer program for executing the step of changing the first threshold when it is determined that the counted number of feature points is equal to or less than the second threshold. A readable recording medium. Let a computer specify a plurality of adjacent pixel regions that are judged to have the same pixel value from a binary image, and extract feature points of the pixel region based on the coordinate values of the pixels of the specified pixel region , A computer-readable recording medium in which a computer program for calculating a feature amount indicating an image feature based on the extracted feature point and determining similarity between images based on the calculated feature amount is recorded In
Causing the computer to count the number of pixels in the identified pixel area;
When determining whether or not the number of pixels in the specified pixel region is greater than or equal to the first threshold, the computer sets a plurality of different first thresholds and determines whether or not each is greater than or equal to the first threshold. A step for judging,
A step of causing the computer to calculate the feature points of the pixel region and to count the number of feature points for each of a plurality of different first threshold values when it is determined that the value is equal to or greater than the first threshold value;
A computer program for executing a step of calculating a feature amount based on a feature point extracted from a pixel region having the largest number of feature points counted is recorded on the computer. Possible recording media. Let a computer specify a plurality of adjacent pixel regions that are judged to have the same pixel value from a binary image, and extract feature points of the pixel region based on the coordinate values of the pixels of the specified pixel region , A computer-readable recording medium in which a computer program for calculating a feature amount indicating an image feature based on the extracted feature point and determining similarity between images based on the calculated feature amount is recorded In
Causing the computer to divide the document image area into a plurality of areas;
Causing the computer to set a first threshold for each divided region;
Causing the computer to count the number of pixels in the specified pixel area in each divided area;
Causing the computer to determine whether or not the number of pixels in the pixel area identified in the divided area is equal to or greater than a first threshold in the divided area;
When the computer determines that the number of pixels in the pixel area specified in the divided area is equal to or greater than the first threshold in the divided area, the feature point of the pixel area is calculated and the number of feature points is calculated. A step of counting;
Causing the computer to determine whether the number of feature points counted is less than or equal to a second threshold;
Causing the computer to calculate the feature amount based on the feature points extracted from the pixel area when it is determined that the number of counted feature points exceeds a second threshold;
And a computer program for executing the step of changing the first threshold when it is determined that the counted number of feature points is equal to or less than the second threshold. A readable recording medium. Let a computer specify a plurality of adjacent pixel regions that are judged to have the same pixel value from a binary image, and extract feature points of the pixel region based on the coordinate values of the pixels of the specified pixel region , A computer-readable recording medium in which a computer program for calculating a feature amount indicating an image feature based on the extracted feature point and determining similarity between images based on the calculated feature amount is recorded In
Causing a computer to determine nearby feature points located within a peripheral region of the extracted feature points;
Causing the computer to count the number of nearby feature points determined;
Causing the computer to determine whether the counted number of feature points is equal to or less than a second threshold;
If the computer determines that the counted number of feature points exceeds a second threshold, calculating the feature amount based on the nearby feature points;
And a computer program for executing the step of changing the range of the peripheral area when it is determined that the score of the counted feature points is equal to or less than a second threshold value. A readable recording medium. Let a computer specify a plurality of adjacent pixel regions that are judged to have the same pixel value from a binary image, and extract feature points of the pixel region based on the coordinate values of the pixels of the specified pixel region , A computer-readable recording medium in which a computer program for calculating a feature amount indicating an image feature based on the extracted feature point and determining similarity between images based on the calculated feature amount is recorded In
Causing the computer to count feature points within a predetermined range when extracting feature points; and
Causing the computer to extract feature points included in the region to be processed from the counted feature points;
Allowing the computer to determine whether the number of extracted feature points is less than a third threshold;
When the computer determines that the number of extracted feature points is equal to or greater than a third threshold, calculating the feature amount based on neighboring feature points located in a peripheral region of the extracted feature points; ,
A computer program for executing a step of changing the range of the peripheral area when it is determined that the extracted score is smaller than a third threshold value is recorded on the computer. recoding media. Let a computer specify a plurality of adjacent pixel regions that are judged to have the same pixel value from a binary image, and extract feature points of the pixel region based on the coordinate values of the pixels of the specified pixel region , A computer-readable recording medium in which a computer program for calculating a feature amount indicating an image feature based on the extracted feature point and determining similarity between images based on the calculated feature amount is recorded In
Causing the computer to count the number of pixels in the identified pixel area;
Causing the computer to determine whether the number of pixels in the identified pixel region is greater than or equal to a first threshold;
If the computer determines that the number of pixels in the specified pixel area is equal to or greater than the first threshold, extracting a feature point of the pixel area;
Causing a computer to determine nearby feature points located within a peripheral region of the extracted feature points;
Causing the computer to count the number of nearby feature points determined;
Causing the computer to determine whether the counted number of feature points is equal to or less than a second threshold;
Causing the computer to calculate the feature amount based on the neighboring feature points when it is determined that the score of the counted feature points exceeds the second threshold;
A computer for causing the computer to execute the step of changing the first threshold value or the first threshold value and the range of the surrounding area when it is determined that the number of feature points counted is equal to or less than the second threshold value. A computer-readable recording medium having a program recorded thereon. Let a computer specify a plurality of adjacent pixel regions that are judged to have the same pixel value from a binary image, and extract feature points of the pixel region based on the coordinate values of the pixels of the specified pixel region , A computer-readable recording medium in which a computer program for calculating a feature amount indicating an image feature based on the extracted feature point and determining similarity between images based on the calculated feature amount is recorded In
Causing the computer to count the number of pixels in the identified pixel area;
When determining whether or not the number of pixels in the specified pixel region is greater than or equal to the first threshold, the computer sets a plurality of different first thresholds and determines whether or not each is greater than or equal to the first threshold. A step for judging,
A step of causing the computer to calculate the feature points of the pixel region and to count the number of feature points for each of a plurality of different first threshold values when it is determined that the value is equal to or greater than the first threshold value;
Causing the computer to extract the feature points of the pixel region having the largest number of feature points counted;
Causing a computer to determine nearby feature points located within a peripheral region of the extracted feature points;
Causing the computer to count the number of nearby feature points determined;
Causing the computer to determine whether the counted number of feature points is equal to or less than a second threshold;
If the computer determines that the counted number of feature points exceeds a second threshold, calculating the feature amount based on the nearby feature points;
And a computer program for executing the step of changing the range of the peripheral area when it is determined that the score of the counted feature points is equal to or less than a second threshold value. A readable recording medium.

Images