JP5003379B2 - Image processing apparatus and program - Google Patents

Description

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

  There is a difference extraction technique that compares two images and extracts a difference between the two images (see, for example, Patent Document 1). The difference extraction technique is used, for example, for detecting a portion where writing is performed when writing is performed later on a document printed on paper.

When two images are compared in this way, for example, an image reading device such as an image scanner reads an image formed on a medium such as paper, so that an image to be compared with an image processing device that performs comparison processing. Enter. In this case, a positional deviation or a positional distortion may occur between the two images to be compared due to a positional deviation of the medium with respect to the reading surface of the image reading apparatus. Therefore, the image processing apparatus performs alignment processing between the two images. As a specific example of the alignment processing, the image processing apparatus, for example, a point indicating a characteristic portion of the image in the image to be compared, such as an end point of a line segment or a representative point of a rectangular area surrounding characters in the image. A process of matching the positions of (feature points) is executed. As a result, the image processing apparatus can remove the influence and compare the two images even when there is a positional shift or positional distortion between the two images.
JP 2004-341914 A

  However, the images to be compared in the above-described examples include various properties such as a document image centered on a character string, a line image centered on a line such as a table or drawing, or a photographic image. . In addition, there are images having different properties for each region, such as a certain region in one image representing a photograph and a certain region representing a sentence. Due to such differences in image properties, for example, feature points cannot be detected accurately, or a sufficient number of feature points cannot be detected for performing the alignment process, and the position can be accurately determined. May not be able to match.

  One of the objects of the present invention is to provide an image processing apparatus and program capable of performing alignment by a processing method according to the properties of an image when aligning images to be compared.

The invention according to claim 1 is an image processing apparatus, a property information acquisition unit that acquires information about properties of two images to be compared, and the two images according to the information about the acquired properties In accordance with the processing method determining means for determining the processing method for performing the alignment of the image and the determined processing method, the feature points included in each of the two images are detected, and the detected feature points substantially coincide with each other As described above, the property information acquisition unit includes a document image including a character string and a line as the property information acquisition unit. Information indicating whether it is a line image including a minute or a pattern image representing a pattern, the type of the image and the type of the feature point correspond to each other, and the processing method determination unit includes: The type of feature points corresponding to the information indicating the type of an image to be serial acquired, the positioning means and selects a type of a feature point to be detected.

  The invention according to claim 2 is the image processing apparatus according to claim 1, further comprising information storage means for storing information associating information indicating the type of the image with a processing method for performing alignment. The processing method determining unit determines the processing method stored in the information storage unit in association with the information indicating the type of the acquired image as the processing method when the two images are aligned. It is characterized by that.

A third aspect of the present invention is the image processing apparatus according to the first or second aspect , wherein the processing method determining means detects the feature point according to the information on the acquired property. A variable to be used in the determination is determined, and the alignment means detects a feature point using the determined variable.

A fourth aspect of the present invention is the image processing apparatus according to any one of the first to third aspects, wherein the processing method determination unit is configured to perform the alignment unit according to information about the acquired property. Determine a plurality of candidates for processing methods when performing alignment and priorities for the candidates for the respective processing methods, and the alignment means determines the plurality of candidates according to the order indicated by the determined priorities. The two images are aligned according to a processing method determined from the above.

The invention according to claim 5 is the image processing device according to claim 1, wherein the property information acquisition unit divides each of the two images into a plurality of regions having different properties, and each of the divided images. Information on the position and property of the region is acquired, and the processing method determining means detects a feature point included in the region for each of the divided regions according to information on the property of the region. And the alignment unit detects the feature points included in each of the divided regions according to the processing method determined by the processing method determination unit for the region, so that the detected feature points substantially match. In addition, the two images are aligned.

A sixth aspect of the present invention is the image processing apparatus according to the fifth aspect , wherein the processing method determining means is characterized in that, out of the plurality of divided areas, according to information relating to properties of the respective areas. Select a detection target region as a target of point detection, and the positioning means detects a feature point included in the selected detection target region, and the two feature points are substantially matched to each other. It is characterized by aligning images.

A seventh aspect of the present invention is the image processing apparatus according to the fifth or sixth aspect , wherein the processing method determining means detects feature points for each of the regions in accordance with information on the properties of the divided regions. The positioning means detects the feature points included in each region according to the order indicated by the determined priorities, and the two feature points are matched so that the detected feature points substantially match. One image is aligned.

The invention according to claim 8 is a property information acquisition means for acquiring information related to the properties of two images to be compared, and a process for performing alignment of the two images according to the information related to the acquired properties According to the processing method determining means for determining the method and the determined processing method, the feature points included in each of the two images are detected, and the two feature images are matched so that the detected feature points substantially coincide with each other. A program for causing a computer to function as an alignment unit that performs alignment , wherein the property information acquisition unit includes a document image including a character string, a line segment as information on the property, Information indicating whether the image is a line image including a pattern image or a pattern image representing a pattern, and the type of the image and the type of the feature point correspond to each other, Serial processing method determination means, the type of feature points corresponding to the information indicating the type of an image to be the acquisition, it said alignment means is a program for selecting a type of a feature point to be detected.

  According to the first and ninth aspects of the invention, it is possible to perform alignment between two images by a processing method according to the properties of the images.

  According to the second aspect of the present invention, it is possible to perform alignment between two images by a processing method according to the type of image.

  According to the third aspect of the present invention, the feature points can be easily detected by detecting the feature points of the type corresponding to the properties of the image.

  According to the fourth aspect of the invention, the feature point detection accuracy can be improved by detecting the feature point using the variable according to the property of the image.

  According to the fifth aspect of the present invention, the alignment processing efficiency can be improved by preferentially using the processing method estimated to be more suitable according to the properties of the image.

  According to the sixth aspect of the present invention, it is possible to perform alignment between two images by a processing method corresponding to the properties of each of a plurality of regions included in the image.

  According to the seventh aspect of the present invention, the feature point detection efficiency is improved by detecting the feature point from the region where the feature point is estimated to be easily detected according to the properties of each of the plurality of regions included in the image. It can be improved.

  According to the eighth aspect of the present invention, feature point detection is performed by preferentially detecting feature points from regions presumed to be easy to detect feature points according to the properties of each of the plurality of regions included in the image. Can improve the efficiency.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings. An image processing apparatus 1 according to an embodiment of the present invention is a personal computer or the like, for example, and as illustrated in FIG. 1, a control unit 11, a storage unit 12, an operation unit 13, a display unit 14, and an interface. Part 15. The image processing apparatus 1 is connected to the image reading apparatus 2.

  The control unit 11 is a CPU or the like, and executes various types of information processing according to programs stored in the storage unit 12. In this embodiment, the control part 11 performs a difference extraction process with respect to two images used as comparison object. A specific example of the process executed by the control unit 11 will be described later.

  The storage unit 12 includes a memory element such as a RAM and a ROM, a hard disk, and the like. The storage unit 12 holds programs executed by the control unit 11 and various data. The storage unit 12 also operates as a work memory for the control unit 11.

  The operation unit 13 is, for example, a keyboard or a mouse, and receives a user instruction operation and outputs the content of the instruction operation to the control unit 11. The display unit 14 is a display device, for example, and displays information according to instructions from the control unit 11.

  The interface unit 15 is a communication interface such as a USB (Universal Serial Bus) standard, and relays data transmission / reception between the control unit 11 and the image reading device 2. The image reading device 2 is an image scanner or the like, reads an image formed on a medium such as paper, and outputs it as electronic data to the image processing device 1.

  Hereinafter, functions realized by the image processing apparatus 1 in the present embodiment will be described. As shown in FIG. 2, the image processing apparatus 1 functionally includes a property information acquisition unit 21, a processing method determination unit 22, a registration processing unit 23, and a difference extraction processing unit 24. The These functions can be realized, for example, when the control unit 11 executes a program stored in the storage unit 12. This program may be provided by being stored in various computer-readable information storage media such as a CD-ROM or a DVD-ROM, or may be provided via communication means such as the Internet.

  The property information acquisition unit 21 first acquires two images (comparison target images) to be compared. And about the acquired comparison object image, the information regarding the property (henceforth property information) is acquired. One image I1 to be compared is an image (original image) before the change, and the other image I2 is an image after the change in which the image I1 is added, corrected, or the like.

  For example, the property information acquisition unit 21 acquires an image read by the image reading device 2 as a comparison target image. That is, the image reading device 2 reads an image formed on a medium such as paper, and outputs the read image data to the image processing device 1. The property information acquisition unit 21 acquires an image output from the image reading device 2 as a comparison target image. In addition, the property information acquisition unit 21 may acquire an image stored in the storage unit 12 or an image transmitted from another information processing apparatus via a communication network as a comparison target image.

  Furthermore, the property information acquisition unit 21 acquires the property information of at least one of the acquired images I1 and I2. The property information is information relating to the characteristics and properties of the image, for example, information indicating the type of the image. Alternatively, when an image includes a plurality of regions having different properties (for example, when a region representing a photograph or a region representing a sentence is mixed in one image), the property information of the image is Such information indicating the position (range) of each region and information regarding the properties of the image in the region may be included.

  The processing method determination unit 22 determines a processing method for aligning the two images I1 and I2 according to the property information acquired by the property information acquisition unit 21. For example, the processing method determination unit 22 detects the type of feature points to be detected and the feature point detection when the alignment processing unit 23 (to be described later) executes processing for detecting feature points of the image from the images I1 and I2. A region to be targeted (detection target region) is determined. Further, the processing method determination unit 22 may determine a variable (processing parameter) used in the alignment process according to the property information.

  In addition, the processing method determination unit 22 determines a plurality of processing method candidates when the alignment processing unit 23 performs alignment, and priorities for the respective processing method candidates according to the property information. Also good. In this case, when the alignment process using the processing method with a higher priority has failed, the processing method determination unit 22 determines a processing method with a lower priority as a new processing method. A specific example of the processing method determined by the processing method determination unit 22 according to the property information will be described in detail later.

  The alignment processing unit 23 performs alignment processing of the two images I1 and I2 according to the processing method determined by the processing method determination unit 22. Specifically, the alignment processing unit 23 detects feature points included in each of the images I1 and I2. Then, geometric transformation is performed on at least one of the two images so that the detected feature points substantially coincide. In the present embodiment, as shown in FIG. 3, the alignment processing unit 23 is functionally a feature point detection unit 23a, a corresponding feature point identification unit 23b, a conversion coefficient calculation unit 23c, an image conversion unit 23d, It is comprised including. With these functions, the alignment processing unit 23 realizes alignment processing between the image I1 and the image I2.

  The feature point detection unit 23a detects feature points included in each of the images I1 and I2. Here, the feature point is a point in each image indicating a characteristic part of the images I1 and I2, and is, for example, an end point of a line segment or a representative point of a rectangular area circumscribing a character in the image. In principle, the feature point detection unit 23a detects feature points with the same processing method for both the image I1 and the image I2.

  The corresponding feature point specifying unit 23b specifies the correspondence between the feature point detected by the feature point detecting unit 23a from the image I1 and the feature point detected from the image I2. That is, the corresponding feature point specifying unit 23b determines a combination of feature points that should be in the same position in each image. Specifically, for example, the corresponding feature point specifying unit 23b finds feature points whose positions in the images match within a predetermined range and whose pixel distribution around the feature points is similar among the detected feature points. , To identify as a corresponding feature point.

  The transformation coefficient calculation unit 23c performs geometric transformation on either or both of the image I1 and the image I2 so that the positions of the feature points identified as corresponding to each other by the corresponding feature point identification unit 23b in the image substantially coincide with each other. A conversion coefficient is calculated. The image conversion unit 23d performs geometric conversion on at least one of the image I1 and the image I2 based on the conversion coefficient calculated by the conversion coefficient calculation unit 23c. Here, the geometric transformation is, for example, affine transformation or the like, and is image processing that translates, rotates, or corrects image distortion. Thereby, the two images I1 and I2 are converted so that the feature points corresponding to each other coincide with each other within a predetermined accuracy range.

  In the above process, the alignment processing unit 23 performs the alignment process according to the processing method determined by the processing method determining unit 22. As a specific example, the feature point detection unit 23 a detects the type of feature point determined by the processing method determination unit 22 from the regions determined as the detection target region by the processing method determination unit 22.

  In addition, when the processing method determination unit 22 has determined a plurality of candidates for the processing method, the alignment processing unit 23 selects the plurality of candidates according to the order indicated by the priority order determined by the processing method determination unit 22. The two images are aligned according to the determined processing method. That is, the alignment processing unit 23 first performs alignment processing by a processing method with a high priority. If the alignment process by the processing method fails, the alignment process is performed by a processing method with a lower priority.

  The difference extraction processing unit 24 compares the image I1 and the image I2 that have been subjected to the alignment processing by the alignment processing unit 23, and extracts differences from each other. As a specific example, the difference extraction processing unit 24 determines whether or not the pixel values of pixels at the same position in each image match each other. And the pixel from which a pixel value differs is extracted as a difference location.

  Hereinafter, some specific examples of the processing executed by the image processing apparatus 1 will be described. First, as a first example, an example in which the property information acquisition unit 21 acquires information indicating the type of comparison target image as property information will be described with reference to the flowchart of FIG.

  First, the property information acquisition unit 21 acquires images I1 and I2 as comparison target images (S1). Then, image type determination is performed on one of the acquired comparison target images (S2).

  Specifically, the property information acquisition unit 21 determines whether the comparison target image is a document image mainly including a character string or a line image mainly including a table or a line segment based on the distribution of pixels included in the comparison target image. Or a pattern image representing a pattern such as a photograph or an illustration. For example, the property information acquisition unit 21 performs image type determination based on the pixel density of pixels whose pixel values are not 0, the degree of convergence (degree of dispersion) of such pixels, and the like. In addition, the property information acquisition unit 21 may execute a labeling process for assigning labels to pixels connected to each other in the comparison target image, and determine the type of the image based on the number of labels given. .

  In addition, when the comparison target image is not a binary image but a grayscale image, a color image, or the like, the property information acquisition unit 21 obtains a binary value obtained as a result of executing a monochromatic process or a binarization process on the comparison target image. You may perform the process mentioned above with respect to the image. Alternatively, the above-described image type determination may be executed based on the pixel density or pixel distribution of black pixels or pixels having a pixel value equal to or greater than a predetermined value. In addition, the property information acquisition unit 21 stores information on image properties input by the user through an instruction operation on the operation unit 13 and information on image properties stored in the storage unit 12 in advance in association with the comparison target image. It may be used for the type determination.

  Here, it is assumed that the image types of the two comparison target images are the same, and the image type determination is performed for only one of the images. However, the property information acquisition unit 21 determines the image for both of the two images. The type may be determined. In this case, when it is determined that the two image types are different from each other, predetermined error processing may be executed.

  Next, the processing method determination unit 22 determines the processing method of the alignment process according to the image type of the comparison target image determined in S2 (S3). Here, specifically, it is assumed that the storage unit 12 stores information that associates information indicating the type of image with a processing method for performing alignment. The processing method determination unit 22 aligns the comparison method image acquired in S1 with the processing method stored in the storage unit 12 in association with the information indicating the type of image acquired by the property information acquisition unit 21 in S2. The processing method is determined.

  Here, a specific example of the processing method determined by the processing method determination unit 22 according to the property information will be described. FIG. 5 is a specific example of information stored in the storage unit 12. In the example of this figure, the type of feature point detected by the feature point detection unit 23a is stored as a processing method of the alignment process in association with information indicating the type of image. Specifically, when the image type is a document image, the processing method determination unit 22 determines to detect a representative point of a rectangular area circumscribing each character included in the image as a feature point. Here, the representative point of the rectangular area is, for example, a rectangular vertex or a center point. If the image type is a line image, end points, intersections, bent points, and the like of line segments are detected as feature points. When the image type is a pattern image, the vertex of the outer frame (outer periphery) surrounding the area representing the pattern such as a photograph is detected as a feature point.

  Further, the processing method determination unit 22 may determine a variable to be used when the feature point detection unit 23a detects a feature point according to the type of the comparison target image. As a specific example, the processing method determination unit 22 determines the threshold value of the pixel value when the feature point detection unit 23a binarizes the comparison target image in order to detect the feature point according to the type of image. Further, the number of feature points to be detected for alignment may be determined according to the type of image.

  Further, the processing method determination unit 22 may determine a detection target region that is a target for the feature point detection unit 23a to detect a feature point according to the type of the comparison target image. For example, when the type of the comparison target image is a design image and the vertex of the outer frame is detected as a feature point as described above, it is estimated that such a feature point is located in the peripheral portion instead of the central portion of the image. Therefore, the processing method determination unit 22 selects such a peripheral region of the image as a detection target region.

  Next, the alignment processing unit 23 detects feature points, specifies correspondences between feature points, and calculates conversion coefficients according to the processing method determined in S3 (S4). That is, first, the feature point detection unit 23a detects the type of feature point selected in S3 from each of the images I1 and I2. Then, the corresponding feature point specifying unit 23b specifies the correspondence between the feature points detected from each image, and the transform coefficient calculating unit 23c is configured so that the corresponding feature points match within a predetermined error range. Then, a conversion coefficient for converting the image is calculated.

  Here, a specific example of the feature point detection process will be described. For example, when the feature point type is an end point of a line segment or an apex of an outer frame surrounding an area representing a pattern, the feature point detection unit 23a applies a pattern to a binary image obtained by binarizing the comparison target image. Feature points are detected by executing known image processing such as matching processing. When the feature point type is a representative point of a rectangular area circumscribing the character, the feature point detection unit 23a detects a pixel group connected to each other by a labeling process, and circumscribes each detected pixel group. Among them, a rectangular area whose width and height are within a predetermined size range and whose pixel density is within a predetermined value range is determined to be a rectangular area representing a character. Then, the position of the feature point is determined according to the position of the rectangular area.

  Subsequently, the alignment processing unit 23 determines whether or not the conversion coefficient satisfying the predetermined condition has been calculated (S5). In the process of S4, when a predetermined number of feature points cannot be detected, when the correspondence between feature points cannot be specified, when a conversion coefficient that matches the corresponding feature points cannot be calculated, or calculated In some cases, such as when the conversion coefficient exceeds a predetermined range, calculation of the conversion coefficient that satisfies the predetermined condition may fail. In such a case, the feature point detection condition (that is, the processing method determined by the processing method determination unit 22) may not be appropriate. Therefore, when the determination in S5 fails, the processing method determination unit 22 newly selects a processing method different from the processing methods tried so far according to the priority order determined according to the property information of the comparison target image. (S6). Then, the alignment processing unit 23 returns to S4 and performs feature point detection again by the new processing method determined in S6. As a specific example, when the type of the image is determined to be a line image and the process of detecting the end point of the line segment as a feature point fails, the processing method determination unit 22 uses the representative point of the rectangular area circumscribing the character as the feature point. Is determined as a new processing method.

  When it is determined that the conversion coefficient has been successfully calculated in S5, the image conversion unit 23d performs geometric conversion on one of the comparison target images using the conversion coefficient calculated in S4 (S7). Then, the difference extraction processing unit 24 compares the comparison target images subjected to the geometric transformation in S7, and executes a difference extraction process (S8). The difference extraction process part 24 produces | generates and outputs the difference image which shows a difference location, for example as an execution result of a difference extraction process. The generated difference image is presented to the user by being displayed on the display unit 14 or printed on a medium such as paper by an image forming apparatus.

  Next, a second example of processing executed by the image processing apparatus 1 will be described based on the flowchart of FIG. In the second example, the property information acquisition unit 21 divides the comparison target image into a plurality of regions, and acquires information on the position and property of each divided region as property information.

  First, the property information acquisition unit 21 acquires images I1 and I2 as comparison target images (S11). Then, an area division process is executed for each acquired comparison target image (S12).

  Specifically, the property information acquisition unit 21 uses, for example, a well-known T / I (Text / Image) separation process to convert each comparison target image into a document area mainly including a character string and a line drawing area including a line image. The image is divided into a plurality of regions having different properties, such as a pattern region including an image such as a photograph or an illustration. In this case, the property information acquisition unit 21 divides the comparison target image into a plurality of regions, and acquires information indicating the image type of the image included in each region as the property information.

  Next, the processing method determination unit 22 determines a processing method for detecting feature points included in each region according to the information regarding the properties of each region obtained by the processing of S12 (S13). As a specific example, the processing method determination unit 22 determines the type of feature point to be detected for each region according to information indicating the image type of the image included in the region. The determination of the feature point type according to the image type can be realized by the same process as the process of S3 in the flow of FIG.

  In this case, the processing method determination unit 22 may select a detection target region as a feature point detection target in accordance with information on the property of each region. For example, the processing method determination unit 22 selects a predetermined image type region (for example, a document region and a line drawing region) as a detection target region, and performs feature point detection for other image type regions (for example, a pattern region). Exclude from

  In addition, the processing method determination unit 22 may determine the priority order of feature point detection for each region according to the information regarding the properties of each region. Specifically, for example, the processing method determination unit 22 determines a document area or a line drawing area as a detection target area with a high priority, and determines a pattern area as a detection target area with a low priority.

  Next, the feature point detection unit 23a detects feature points from the two comparison target images in accordance with the processing method determined in S13 (S14). Specifically, the feature point detection unit 23a detects the feature points included in each region divided in S12 according to the processing method determined by the processing method determination unit 22 for the region. At this time, feature points are detected for the region selected as the detection target region, and feature points are not detected for other regions. In addition, when the priority order of feature point detection for each region is determined, the feature points included in each region are detected according to the order indicated by the priority order. In this case, for example, the feature point detection unit 23a sequentially detects feature points for each region until a predetermined number of feature points are detected.

  Subsequently, the corresponding feature point specifying unit 23b and the conversion coefficient calculating unit 23c specify the correspondence between the feature points detected from the images in S14, and calculate the conversion coefficient so that the corresponding feature points are substantially matched. Perform (S15). Then, the image conversion unit 23d performs geometric conversion on one of the comparison target images using the conversion coefficient calculated in S15 (S16). Further, the difference extraction processing unit 24 compares the comparison target images subjected to the geometric transformation in S16, and executes a difference extraction process (S17). These processes are realized by a process similar to the example of the flow of FIG.

  Through the processing described above, the image processing apparatus 1 performs alignment between two images by a processing method according to the properties of the image to be compared. As a result, alignment is performed with high accuracy in accordance with the properties of the image.

  The embodiment of the present invention is not limited to the above-described embodiment. For example, the image processing apparatus 1 is not limited to the difference extraction process, and may perform the above-described alignment process as a pre-process for various processes for comparing two images. In the above description, the image processing apparatus 1 acquires information indicating the type of image, information indicating the type of image for each area, and the like as property information. It is also possible to acquire various types of information indicating the characteristics and to determine the processing method accordingly. Also, the processing method determined by the processing method determination unit 22 is not limited to the type of feature point to be detected and the value of the variable, and the processing method used for alignment may be determined from other processing methods.

  The image processing apparatus 1 may acquire two image sequences including a plurality of pages representing a series of documents, and may sequentially compare the images included in each image sequence. In such a case, for the first images in the image sequence, the property information is acquired from the pixel distribution or the like of the comparison target image as described above. However, for the second and subsequent images, the property information acquisition unit 21 estimates that the property information is the same as the image of the previous page for which property information has already been acquired, and a processing method according to the estimated property information The alignment process may be executed by This simplifies the processing when, for example, document images are continuously included in each image sequence.

  In the above description, the processing method is determined and the alignment process such as feature point detection is performed on the two comparison target images before and after the change at the same time. However, for example, with respect to the original image I1 that is the original, the processing such as feature point detection may be executed in advance by a processing method corresponding to the property information, and the execution result may be stored in the storage unit 12. In this case, when the image processing apparatus 1 acquires the changed image I2, the image processing apparatus 1 performs processing such as feature point detection on the image I2 in the same processing method as the processing method executed on the image I1 before the change. Execute. Then, alignment is performed using the execution result and the execution result of the image I1 before change stored in the storage unit 12. In addition, for the image I1 before the change, an execution result obtained by executing processing such as feature point detection by each of a plurality of processing methods according to a predetermined priority order may be stored. In this case, even if the image processing apparatus 1 cannot detect a sufficient number of feature points with a certain processing method for the image I2 after the change, the image processing device 1 performs the detection of the feature points with another processing method and executes the detection. Using the result and the execution result for the image I1 before change stored in the storage unit 12, alignment is executed.

1 is a block diagram illustrating a schematic configuration of an image processing apparatus according to an embodiment of the present invention. It is a functional block diagram showing the function of the image processing apparatus which concerns on embodiment of this invention. It is a functional block diagram showing the function of a position alignment process part. It is a flowchart which shows an example of the process which the image processing apparatus which concerns on embodiment of this invention performs. It is explanatory drawing which shows an example of the information memorize | stored in the image processing apparatus which concerns on embodiment of this invention. It is a flowchart which shows another example of the process which the image processing apparatus which concerns on embodiment of this invention performs.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 Image processing apparatus, 2 Image reading apparatus, 11 Control part, 12 Memory | storage part, 13 Operation part, 14 Display part, 15 Interface part, 21 Property information acquisition part, 22 Processing method determination part, 23 Position alignment processing part, 24 Difference Extraction processing unit.

Claims (8)

Property information acquisition means for acquiring information about the properties of the two images to be compared;
A processing method determining means for determining a processing method when performing alignment of the two images according to the information on the acquired property;
In accordance with the determined processing method, a feature point included in each of the two images is detected, and a registration unit that aligns the two images so that the detected feature points substantially coincide with each other.
Have
The property information acquisition means includes information indicating whether the type of the two images is a document image including a character string, a line image including a line segment, or a pattern image representing a pattern as the information regarding the property. Acquired,
The image type and the feature point type correspond to each other,
The image processing apparatus, wherein the processing method determination unit selects a feature point type corresponding to information indicating the type of the acquired image as a feature point type detected by the alignment unit . An information storage unit that stores information associating information indicating the type of the image with a processing method for performing alignment;
The processing method determining means determines a processing method stored in the information storage means in association with information indicating the type of the acquired image as a processing method for performing alignment of the two images. The image processing apparatus according to claim 1. The processing method determining means determines a variable to be used when the positioning means detects a feature point according to information on the acquired property,
It said alignment means, an image processing apparatus according to claim 1 or 2, characterized in that to detect the feature points by using the variables the determined. The processing method determining unit determines a plurality of processing method candidates when the alignment unit performs alignment and a priority order for each processing method candidate according to the acquired property information. And
4. The alignment of the two images according to claim 1, wherein the alignment unit aligns the two images according to a processing method determined from the plurality of candidates according to an order indicated by the determined priority order. 5 . The image processing apparatus according to any one of claims. The property information acquisition unit divides each of the two images into a plurality of regions having different properties from each other, acquires information on the position and property of each of the divided regions,
The processing method determining means determines, for each of the divided regions, a processing method for detecting feature points included in the region according to information on the properties of the region,
The alignment means detects the feature points included in each of the divided areas according to the processing method determined by the processing method determination means for the area, and the two feature points are substantially matched with each other. The image processing apparatus according to claim 1, wherein alignment of two images is performed. The processing method determining means selects a detection target region as a target for feature point detection from the plurality of divided regions, according to information on the properties of each region,
The alignment unit detects a feature point included in the selected detection target region, and aligns the two images so that the detected feature point substantially matches. 5. The image processing apparatus according to 5 . The processing method determining means determines priority order of feature point detection for each area according to information on the properties of each divided area,
The alignment unit detects feature points included in each region according to the order indicated by the determined priority order, and aligns the two images so that the detected feature points substantially coincide with each other. The image processing apparatus according to claim 5 or 6, Property information acquisition means for acquiring information on properties of two images to be compared;
Included in each of the two images in accordance with the processing method determining means for determining a processing method for aligning the two images according to the information on the acquired properties, and according to the determined processing method Alignment means for detecting feature points and aligning the two images so that the detected feature points substantially match;
A program for causing a computer to function as,
The property information acquisition means includes information indicating whether the type of the two images is a document image including a character string, a line image including a line segment, or a pattern image representing a pattern as the information regarding the property. Acquired,
The image type and the feature point type correspond to each other,
The processing method deciding means selects a feature point type corresponding to information indicating the type of the acquired image as a feature point type detected by the alignment means .

Images