JP7750255B2 - Image processing device - Google Patents

Description

The present invention relates to an image processing device.

An image processing device is known that compares a pre-application image with a post-application image and extracts additional information from the post-application image. This image processing device searches for white areas in the image and divides the entire image into multiple regions based on these white areas. The pre-application image and the post-application image are then aligned relative to each other for each divided region, and after alignment, the additional information is extracted from the post-application image (see, for example, Patent Document 1).

Japanese Patent Application Laid-Open No. 2007-335920

When the relative positioning of the pre-addition image and the post-addition image is performed for each divided region (hereinafter referred to as a partial region), the correspondence between the pre-addition image and the post-addition image can be determined based on the results of comparing the feature amounts of each partial region. For example, this feature amount is calculated based on the pixels contained in the region of interest, which defines the range for calculating the feature amount, as it moves over the specified partial region for which the feature amount is to be calculated.

However, when the area of interest moves over a specified partial area, it may overlap another partial area adjacent to the specified partial area. As a result, the feature values are calculated based on the pixels of the specified partial area and the pixels of the other partial area, which may reduce the accuracy of the feature values. If an attempt is made to identify the correspondence between the pre-application image and the post-application image based on such low-accuracy feature values, it may not be possible to accurately identify the corresponding partial areas.

The present invention therefore aims to provide an image processing device that improves the accuracy of identifying corresponding partial regions.

The image processing device according to the present invention includes a control unit that executes the following processes: dividing first drawing data from a first region representing one page of the first drawing data into a plurality of first partial regions; dividing second drawing data from a second region representing one page of the second drawing data into a plurality of second partial regions; generating a first extended region in which a first predetermined region of a uniform color is set on the periphery of one of the specific first partial regions when calculating a first feature value for one of two specific adjacent first partial regions among the plurality of first partial regions if the other specific first partial region interferes with the calculation of the first feature value; generating a second extended region in which a second predetermined region of a uniform color is set on the periphery of one of the specific second partial regions when calculating a second feature value for one of two specific adjacent second partial regions among the plurality of second partial regions when the other specific second partial region interferes with the calculation of the second feature value; and comparing the first extended region with the second extended region and identifying the second extended region that corresponds to the first extended region based on the comparison result.

In the above configuration, the control unit may execute processing to generate the first expanded region and the second expanded region when a first positional relationship between one and the other of the specific first partial regions and a second positional relationship between one and the other of the specific second partial regions are relatively different.

In the above configuration, the control unit may execute processing to identify the second expansion region corresponding to the first expansion region based on first feature points possessed by the first expansion region, second feature points possessed by the second expansion region, and a predetermined feature point matching method using machine learning.

In the above configuration, the control unit may execute processing to generate the first extended region in which the first predetermined region is set to a size that avoids interference with the calculation of the first feature amount by the other of the specific first partial regions, and to generate the second extended region in which the second predetermined region is set to a size that avoids interference with the calculation of the second feature amount by the other of the specific second partial regions.

In the above configuration, both the first specified area and the second specified area may be blank space.

This invention can improve the accuracy of identifying corresponding partial regions.

1 is an example of a hardware configuration of an image processing device. 1 is an example of a functional configuration of an image processing device. 1A is an example of first medium data, and FIG. 1B is an example of second medium data. 10 is a flowchart illustrating an example of the operation of a control unit. 1A is a diagram illustrating a case where different feature amounts are calculated, FIG. 1B is a diagram illustrating generation of an extension region, and FIG. 1C is a diagram illustrating comparison between extension regions. 1A is an example of pairing of continuous regions based on the number of feature amounts, and FIG. 1B is an example of pairing of continuous regions based on the distance of feature amounts. 1A is an example of a difference image, and FIG. 1B is another example of a difference image.

The following describes an embodiment of the present invention with reference to the drawings.

First, the hardware configuration of the image processing device 100 will be described with reference to Figure 1. The image processing device 100 includes a CPU (Central Processing Unit) 100A as a processor, and RAM (Random Access Memory) 100B and ROM (Read Only Memory) 100C as memories. The image processing device 100 also includes a network I/F (Interface) 100D and HDD (Hard Disk Drive) 100E. An SSD (Solid State Drive) may be used instead of the HDD (Hard Disk Drive) 100E.

The image processing device 100 may include at least one of an input I/F 100F, an output I/F 100G, an input/output I/F 100H, and a drive device 100I, as necessary. The CPU 100A to the drive device 100I are connected to each other via an internal bus 100J. In other words, the image processing device 100 can be realized by a computer.

An input device 11 is connected to the input I/F 100F. Examples of the input device 11 include a keyboard, mouse, and touch panel. A display device 12 is connected to the output I/F 100G. Examples of the display device 12 include an LCD display. A semiconductor memory 13 is connected to the input/output I/F 100H. Examples of the semiconductor memory 13 include a USB (Universal Serial Bus) memory and a flash memory. The input/output I/F 100H reads a predetermined program stored in the semiconductor memory 13. The input I/F 100F and the input/output I/F 100H each include a USB port, for example. The output I/F 100G includes a display port, for example.

A portable recording medium 14 is inserted into the drive device 100I. Examples of the portable recording medium 14 include removable disks such as CD (Compact Disc)-ROM and DVD (Digital Versatile Disc). The drive device 100I reads a specific program recorded on the portable recording medium 14. The network I/F 100D includes, for example, a LAN (Local Area Network) port and a communication circuit.

A predetermined program stored in at least one of ROM 100C, HDD 100E, and semiconductor memory 13 is temporarily stored in RAM 100B by CPU 100A. A predetermined program recorded on portable recording medium 14 is temporarily stored in RAM 100B by CPU 100A. By executing the stored predetermined program, CPU 100A realizes various functions described below and executes various processes described below. The predetermined program may be one that corresponds to the flowchart described below.

The functional configuration of the image processing device 100 will be described with reference to Figures 2 and 3. Note that Figure 2 shows the main functions of the image processing device 100.

As shown in FIG. 2, the image processing device 100 includes a memory unit 110, a control unit 120, an input unit 130, and an output unit 140. The memory unit 110 can be realized by the RAM 100B or HDD 100E described above. The control unit 120 can be realized by the CPU 100A described above. The input unit 130 can be realized by the input I/F 100F described above. The output unit 140 can be realized by the output I/F 100G described above. Therefore, the memory unit 110, control unit 120, input unit 130, and output unit 140 are connected to each other.

Here, the storage unit 110 includes a first medium storage unit 111, a second medium storage unit 112, and a display drawing storage unit 113. At least one of the first medium storage unit 111, the second medium storage unit 112, and the display drawing storage unit 113 may be provided in an information processing device (not shown) different from the image processing device 100. In this case, the image processing device 100 may access the information processing device and refer to the stored contents of the first medium storage unit 111, the second medium storage unit 112, etc.

The first medium storage unit 111 stores first medium data. The first medium data is a data file or electronic file of an information transmission medium. The information transmission medium includes documents such as reports and instructions. In this embodiment, as shown in FIG. 3(a), the first medium data 50 is a document created on multiple pages, but it may also be a document created on a single page. When the first medium data 50 is a document created on multiple pages, depending on the document, it may include multiple pieces of drawing data 51, 52, or one piece of drawing data 53. The drawing data 51, 52, 53 are, for example, blueprints of finished products or parts. The shapes of the drawing data 51, 52, 53 are different from each other.

The second medium storage unit 112 stores the second medium data. Like the first medium data 50, the second medium data is a data file or electronic file of an information transmission medium. In this embodiment, as shown in FIG. 3(b), the second medium data 60 is also a document created with multiple pages, but it may also be a document created with a single page. The number of pages in the second medium data 60 may be the same as the number of pages in the first medium data 50, or may be different. When the second medium data 60 is a document created with multiple pages, depending on the document, it may include one piece of drawing data 61, or multiple pieces of drawing data 62, 63. The drawing data 61, 62, 63 include, for example, blueprints of finished products or parts. The shapes of the drawing data 61, 62, 63 are different from each other.

The above-mentioned first medium data 50 is stored in the first medium storage unit 111 based on operations on the input device 11. The second medium data 60 is similarly stored in the second medium storage unit 112. In this embodiment, the first medium data 50 represents the source of comparison for the second medium data 60, and the second medium data 60 represents the destination or comparison target for the first medium data 50. Therefore, hereinafter, the drawing data 51, 52, 53 will be referred to as base drawings 51, 52, 53 serving as the source of comparison, and the drawing data 61, 62, 63 will be referred to as new drawings 61, 62, 63 serving as the destination or comparison target.

The display drawing storage unit 113 stores images of drawings to be displayed on the display device 12. As will be described in detail below, the display drawing storage unit 113 stores an image of a drawing in which a base drawing image corresponding to the above-mentioned base drawings 51 and 52 is superimposed on a new drawing image corresponding to any two of the new drawings 61, 62, and 63 corresponding to these base drawing images, with portions corresponding to the differences expressed in different colors. The display drawing storage unit 113 also stores an image of a drawing in which a base drawing image corresponding to the above-mentioned base drawing 53 is superimposed on a new drawing image corresponding to any one of the new drawings 61, 62, and 63 corresponding to this base drawing image, with portions corresponding to the differences expressed in different colors. The base drawing image is an example of first drawing data, and the new drawing image is an example of second drawing data. Both the base drawing image and the new drawing image are image data.

Returning to Figure 2, the control unit 120 includes an image generation unit 121, an image processing unit 122, and a display unit 123.

The image generation unit 121 acquires the first medium data 50 from the first medium storage unit 111 and images the first medium data 50. That is, the image generation unit 121 generates a first medium image representing the first medium data 50 from the acquired first medium data 50. If the first medium data 50 is a document created with multiple pages, the first medium image also includes multiple page images. The multiple page images included in the first medium image include the base drawing image described above.

Similarly, the image generation unit 121 acquires the second medium data 60 from the second medium storage unit 112 and images the second medium data 60. That is, the image generation unit 121 generates a second medium image representing the second medium data 60 from the acquired second medium data 60. If the second medium data 60 is a document created with multiple pages, the second medium image also includes multiple page images. The multiple page images included in the second medium image include the new drawing image described above.

The image processing unit 122 performs various image processing operations on the first and second medium images, as well as the base drawing image and new drawing image. Details of the image processing unit 122 will be described later. For example, the image processing unit 122 compares multiple page images included in the first medium image with multiple page images included in the second medium image to identify page images that are similar to each other. The image processing unit 122 also divides the area representing the page image into multiple partial areas and calculates a first feature value for one of two specific partial areas adjacent to each other. If the other specific partial area interferes with the calculation of the first feature value, the image processing unit 122 generates an extended area in which a predetermined area of uniform color is set on the periphery of one of the specific partial areas. The predetermined uniform color area is preferably a margin of the same color as the background, but may be a color other than white (e.g., red or blue). Furthermore, the image processing unit 122 identifies corresponding extended areas based on a known feature point matching method using machine learning. The shape of the extended area may be rectangular or a shape other than rectangular (e.g., circular). Additionally, the image processing unit 122 converts each of the base drawing image and the new drawing image to monochrome (specifically, monochrome binarization), and assigns different colors to each of the base drawing image and the new drawing image.

As a known feature point matching method, for example, the following documents 1 and 2 can be referred to.
Reference 1: PF Alcantarilla, A. Bartoli, and AJ Davison. “KAZE features”, ECCV, (2012), pp.214?227.
Reference 2: P. F, Alcantarilla, J. Nuevo, A. Bartoli, “Fast Explicit Diffusion for Accelerated Features in Nonlinear Scale Spaces”, BMVC, (2013), pp.13.1-13.11.

The display unit 123 acquires an image of the drawing stored in the display drawing storage unit 113 and displays the acquired drawing on the display device 12. As described above, the display drawing storage unit 113 stores, for example, an image of the drawing in which parts corresponding to the differences between the base drawing image and the new drawing image are expressed in different colors. Therefore, the display unit 123 acquires and displays this drawing.

Next, the processing performed by the control unit 120 will be described with reference to Figures 4 to 7.

First, as shown in FIG. 4, the image generation unit 121 creates an image of the medium data (step S1). More specifically, the image generation unit 121 acquires first medium data 50 from the first medium storage unit 111 and generates a first medium image. The image generation unit 121 also acquires second medium data 60 from the second medium storage unit 112 and generates a second medium image.

Once the medium data has been imaged, the image processing unit 122 identifies similar pages (step S2). Specifically, the image processing unit 122 compares the multiple page images included in the first medium image with the page images included in the second medium image, page by page, to calculate the similarity between the page images. After calculating the similarity, the image processing unit 122 identifies the page images with the highest calculated similarity as similar pages. This allows similar pages to be uniquely identified. In this embodiment, for example, page images containing the characters "Figure 1" are identified as similar pages (see Figures 3(a) and (b)).

When similar pages are identified, the image processing unit 122 divides each region into partial regions (step S3). That is, the image processing unit 122 divides each of the two page images identified as similar pages into multiple partial regions from a single region corresponding to the entire page image. After performing region division, for each of the two page images identified as similar pages, the image processing unit 122 retains one of two adjacent specific partial regions and deletes the other specific partial region (step S4).

Explaining in more detail, the image processing unit 122 first identifies two specific adjacent partial areas among multiple partial areas. For example, as shown in FIG. 5(a), the image processing unit 122 identifies partial area 71, which is one of the two specific partial areas resulting from the comparison source page image, and partial area 72, which is the other of the two specific partial areas. Partial area 71 includes base drawing 52, and partial area 72 includes base drawing 51. Similarly, the image processing unit 122 identifies partial area 81, which is one of the two specific partial areas resulting from the comparison target page image, and partial area 82, which is the other of the two specific partial areas. Partial area 81 includes new drawing 63, and partial area 82 includes new drawing 62.

Here, when the image processing unit 122 moves a circular region of interest IR, which defines the range for calculating the feature amounts, over each of the partial regions 71 and 81 and calculates the feature amounts based on the pixels included in the region of interest IR, different feature amounts may be calculated. For example, when the region of interest IR moves over partial region 71, part of partial region 72 may be included in the region of interest IR. In this case, partial region 72 interferes with the calculation of the feature amounts of partial region 71. In other words, the feature amounts calculated by the image processing unit 122 include not only the feature amounts of partial region 71 but also the feature amounts of part of partial region 72.

Similarly, when the region of interest IR moves over the partial region 81, part of the partial region 82 may be included in the region of interest IR. In this case, the partial region 82 interferes with the calculation of the feature values of the partial region 81. That is, the feature values calculated by the image processing unit 122 include not only the feature values of the partial region 81 but also the feature values of part of the partial region 82. In particular, as shown in FIG. 5(a), if the positional relationship between the partial regions 71 and 72 and the positional relationship between the partial regions 81 and 82 are relatively different, the feature values calculated by the image processing unit 122 may differ significantly. For example, if these positional relationships are relatively different, the shape of the part of the partial region 72 included in the region of interest IR will differ from the shape of the part of the partial region 82 included in the region of interest IR. In the example of FIG. 5(a), the corners of the partial region 82 are included in the region of interest IR, but the corners of the partial region 72 are not included in the region of interest IR. As a result, the feature amounts of some of the partial regions 72 and 82 differ, and as a result, the feature amounts calculated by the image processing unit 122 differ significantly.

In such a case, as shown in FIG. 5(b), the image processing unit 122 leaves partial region 71 and deletes partial region 72. Also, although not shown, as in the case of partial regions 71 and 72, the image processing unit 122 leaves partial region 81 and deletes partial region 82.

Once the processing of step S4 is completed, the image processing unit 122 sets a predetermined region and generates an extended region (step S5). For example, as shown in FIG. 5(b), the image processing unit 122 generates extended region 92 by setting a predetermined region 91 of a uniform color on the outer periphery of partial region 71. Furthermore, as in the case of partial region 71, the image processing unit 122 generates extended region 96 by setting a predetermined region 95 of a uniform color on the outer periphery of partial region 81.

Predetermined region 91 is larger than partial region 71 and is sized to avoid interference with feature calculation due to partial region 72. Furthermore, predetermined region 95 is larger than partial region 81 and is sized to avoid interference with feature calculation due to partial region 82. As a result, as shown in FIG. 5(b), even if the region of interest IR moves over extended region 92, interference due to partial region 72 can be avoided. Although not shown, even if the region of interest IR moves over extended region 96, interference due to partial region 82 can be avoided.

Once the processing of step S5 is completed, the image processing unit 122 uses the feature point matching method described above to identify the correspondence between the extended regions (step S6). More specifically, the image processing unit 122 uses the feature point matching method to identify the correspondence between the comparison source extended region 92 and the comparison target extended region 96, as shown in FIG. 5(c). Identifying the correspondence includes a first stage and a second stage. The order of the first and second stages is not particularly limited, and either the first stage or the second stage can be performed first.

First, the first stage will be described. When the comparison source extended region has multiple points (e.g., pixels, or local or minute regions containing multiple pixels as elements) A and X, and the comparison target extended region has multiple points B, C, Y, etc., the image processing unit 122 calculates the distance L(A, B) between points A and B and the distance L(A, C) between points A and C in the first stage. Each of points A, B, C, X, Y, etc. has multiple components (e.g., lightness, saturation, luminance, average pixel value (or RGB value), variance of pixel values, histogram of pixel values, etc.) as feature quantities. Therefore, points A and X may be referred to as first feature points, and points B, C, and Y may be referred to as second feature points. For example, point X has lightness as component _x1 , saturation as component _x2 , ..., and a histogram of pixel values as component _xn . Points A, B, C, and Y are basically the same as point X, and detailed description thereof will be omitted.

As a result, the image processing unit 122 can calculate, for example, the distance L(X, Y) using the following calculation formula: Similarly, the image processing unit 122 can also calculate the distance L(A, B) and the distance (A, C) using this calculation formula. Based on the distance L(X, Y) and the distance L(A, B), the image processing unit 122 can determine the similarity between the feature amount of point X and the feature amount of point Y, and the similarity between the feature amount of point A and the feature amount of point B.
<Calculation formula>
Distance L (X, Y) = {(x ₁ - _{y 1} ) ² + (x ₂ - _{y 2} ) ² +...+(x _n - _{y n} ) ² }} ^0.5

Here, the image processing unit 122 uses the number of pairs of elements in the extended area that have unique features (hereinafter referred to as element pairs) to select candidates for extended areas (hereinafter referred to as pairing candidates) to identify correspondences (to be paired). Specifically, the image processing unit 122 selects as pairing candidates the extended areas that have the largest number of point pairs with matching features of points within the extended areas. The image processing unit 122 determines that the features match when the distance L between two points is the smallest compared to the distance L between other two points.

An element pair having a unique feature is required to satisfy the following inequality, based on, for example, the distance L(A, B) between point A and point B, which is the shortest distance and is a candidate pair for point A, and the distance L(A, C) between point A and point C, which is the next shortest distance after point B, which is a candidate pair for point A. If the distance L(A, B) between point A in the source extended region and point B in the target extended region is the shortest, the feature values of points A and B are very similar, and the image processing unit 122 can determine that the feature values match. The first threshold can be set to a positive value less than 1.0, such as 0.5 or 0.6.
<Inequalities>
Distance L(A, B)/Distance L(A, C)<first threshold

As a result, as shown in FIG. 6(a), the image processing unit 122 can select pairing candidates for multiple corresponding extension areas (referred to as areas in FIG. 6(a)) #1 to #5. For example, the extension area candidates corresponding to extension area #3 of the base drawing image are extension area #2 and extension area #3 of the new drawing image. Here, the number of point pairs where the features of points in extension area #3 of the base drawing image match those of points in extension area #2 of the new drawing image is 80. On the other hand, the number of point pairs where the features of points in extension area #3 of the base drawing image match those of points in extension area #3 of the new drawing image is 40. Therefore, the image processing unit 122 determines extension area #3 of the base drawing image and extension area #2 of the new drawing image, which have the largest number of point pairs, as extension area pairing candidates.

Next, the second stage will be described. In the second stage, the image processing unit 122 selects pairing candidates for the extended region that identify correspondences using the shortest distances between the feature quantities of the elements in the extended region. For example, the image processing unit 122 selects several to a dozen (e.g., 5, 10, 15, etc.) point pairs in ascending order of distance L, and selects pairing candidates if the average value obtained by dividing the sum of the distances L of the selected point pairs by the number of points is less than the second threshold.

Here, if the second threshold is set to several tens of values, such as a threshold of "30" or a threshold of "50," the image processing unit 122 can select pairing candidates for multiple corresponding extension areas (referred to as areas in FIG. 6(b)) #1 to #5, as shown in FIG. 6(b). For example, the extension area candidates corresponding to extension area #3 of the base drawing image are extension area #2 and extension area #3 of the new drawing image.

Here, the average distance L between a point in extension area #3 of the base drawing image and a point in extension area #2 of the new drawing image is "10." On the other hand, the average distance L between a point in extension area #3 of the base drawing image and a point in extension area #3 of the new drawing image is "13." Therefore, the image processing unit 122 determines extension area #3 of the base drawing image and extension area #2 of the new drawing image, which have the shortest average value of distance L, as pairing candidates for extension areas. Note that the image processing unit 122 may select pairing candidates using the above-mentioned sum and a fifth threshold value that is greater than the above-mentioned second threshold value, without using the average value.

The image processing unit 122 ultimately determines the pair of the comparison source and comparison destination extended areas based on the pairing candidates determined in the first stage and the pairing candidates determined in the second stage. Therefore, as shown in Figures 6(a) and 6(b), the image processing unit 122 pairs, for example, extended area #1 of the base drawing image with extended area #1 of the new drawing image as extended areas having a corresponding relationship.

As shown in Figure 6(a), the image processing unit 122 selects extended area #5 of the base drawing image and extended area #5 of the new drawing image as pairing candidates, but as shown in Figure 6(b), it excludes extended area #5 of the base drawing image and extended area #5 of the new drawing image from the selection of pairing candidates. This is because the average value of the distance L between extended area #5 of the base drawing image and extended area #5 of the new drawing image exceeds the second threshold. In this case, the image processing unit 122 may determine that extended area #5 has been deleted from the base drawing image and added to the new drawing image.

In this way, by identifying correspondences that include the first and second stages, the image processing unit 122 pairs the extended regions with each other. In other words, the image processing unit 122 avoids a situation where a partial region included in the new drawing image to be compared is not determined as a pairing target due to low-precision features compared to a partial region included in the base drawing image to be compared. In this way, by recognizing and generating extended regions, the image processing unit 122 can accurately and uniquely identify correspondences between extended regions, and can accurately identify correspondences between partial regions. Note that while extended regions for drawings have been described as an example, the image processing unit 122 may also identify and generate extended regions for text in the same way as for drawings, and identify correspondences.

Once the correspondence has been identified, the image processing unit 122 moves one of the extended regions for which the correspondence has been identified to the other (step S7), aligning the positions of the base drawing image and the new drawing image (step S8). Once the positions of the base drawing image and the new drawing image have been aligned, the image processing unit 122 converts the base drawing image to monochrome and assigns green alone. Green is one of the three primary colors of light, and can be expressed as (R, G, B) = (0, 255, 0) in the RGB format, which classifies brightness into 256 levels. In the 256-level brightness classification, the numerical data "255" represents the brightest brightness. By converting the base drawing image to monochrome and assigning green alone, the base drawing image can be expressed in green.

When the base drawing image is converted to monochrome and green is added alone, the image processing unit 122 converts the new drawing image to monochrome and red is added alone. Red is one of the three primary colors of light, and can be expressed in RGB color representation as (R, G, B) = (255, 0, 0). By converting the new drawing image to monochrome and adding red alone, the new drawing image can be expressed in red. Note that green may be added to the new drawing image and red may be added to the base drawing image.

When the new drawing image is converted to monochrome and red is added alone, the image processing unit 122 overlays the base drawing image and new drawing image on a blue background. Blue is one of the three primary colors of light, and can be expressed as (R, G, B) = (0, 0, 255) in RGB color representation. The background is an example of specified data. When the base drawing image and new drawing image are overlaid on a blue background, the result is (R, G, B) = (255, 255, 255), which allows white to be expressed.

In this way, the portion of the white background that contains only the green base drawing image can be expressed as magenta, including pink, because (R,G,B) = (255,255,255) - (0,255,0) = (255,0,255). Similarly, the portion of the white background that contains only the red new drawing image can be expressed as cyan, including light blue, because (R,G,B) = (255,255,255) - (255,0,0) = (0,255,255). Furthermore, the overlapping portion of the white background containing the green base drawing image and the red new drawing image can be expressed as blue, because (R,G,B) = (255,255,255) - (0,255,0) - (255,0,0) = (0,0,255).

When the base drawing image and the new drawing image are superimposed on a blue background, the image processing unit 122 stores an image of such a drawing in the display drawing storage unit 113. Once the drawing image has been stored in the display drawing storage unit 113, the display unit 123 displays the drawing image on the display device 12 (step S9), and the processing ends. As a result, as shown in Figures 7(a) and (b), the base drawing image and the new drawing image are superimposed, and an image of the drawing, partially expressed in different colors, appears on the display device 12.

For example, as shown in FIG. 7(a), the differences between the base drawing image corresponding to base drawing 51 and the new drawing image corresponding to new drawing 62, and the differences between the base drawing image corresponding to base drawing 52 and the new drawing image corresponding to new drawing 63, appear on the display device 12 as difference images P1 and P2 of different colors. In this case, difference image P1, which is part of the base drawing image, is displayed in magenta, and difference image P2, which is part of the new drawing image, is displayed in cyan. Then, superimposed image P3, which is the superimposed portion of the base drawing image and the new drawing image, is displayed in blue.

Also, as shown in Figure 7(b), the differences between the base drawing image corresponding to the base drawing 53 and the new drawing image corresponding to the new drawing 61 appear on the display device 12 as a difference image Q1 of a different color. In this case, the difference image Q1, which is part of the base drawing image, is displayed in magenta, but since the entire new drawing image is superimposed on the base drawing image, the new drawing image is displayed as a blended image in blue as a superimposed image Q3.

In this way, in the image processing device 100 according to this embodiment, for each of two page images identified as similar pages, the image processing unit 122 calculates the feature amounts of two specific adjacent partial regions 71, 81. If the partial regions 72, 82 interfere with the feature amount calculation, the image processing unit 122 generates extended regions 92, 96 in which predetermined regions 91, 95 of uniform color are set around the periphery of the partial regions 71, 81, and identifies the correspondence between the extended regions 92, 96. This prevents the low accuracy of identifying the partial regions 71, 81 when the extended regions 92, 96 are not generated. As a result, the accuracy of identifying the partial region 71 included in the extended region 92 of the comparison source and the partial region 81 included in the extended region 96 of the comparison target can be improved.

Although the preferred embodiment of the present invention has been described in detail above, the present invention is not limited to such specific embodiment, and various modifications and variations are possible within the scope of the gist of the present invention as set forth in the claims.

50 First medium data 51, 52, 53 Base drawing 60 Second medium data 61, 62, 63 New drawing 71, 72, 81, 82 Partial area 91, 95 Predetermined area 92, 96 Extended area 100 Image processing device 120 Control unit IR Area of interest

Claims (5)

A process of dividing the first drawing data into a plurality of first partial regions from a first region representing an image of one page of the first drawing data, and dividing the second drawing data into a plurality of second partial regions from a second region representing an image of one page of the second drawing data;
a process of generating a first extended region in which a first predetermined region of a uniform color is set on an outer periphery of one of the specific first partial regions when a first feature amount of one of two specific first partial regions adjacent to each other among the plurality of first partial regions is calculated and the other specific first partial region interferes with the calculation of the first feature amount;
a process of generating a second extended region in which a second predetermined region of a uniform color is set on an outer periphery of one of the specific second partial regions when calculating a second feature amount of one of the two specific second partial regions adjacent to each other among the plurality of second partial regions and the other specific second partial region interferes with the calculation of the second feature amount;
a process of comparing the first extension region with the second extension region and identifying the second extension region corresponding to the first extension region based on a comparison result;
An image processing device having a control unit that executes the above. the control unit generates the first extended region and the second extended region when a first positional relationship between one and the other of the specific first partial regions and a second positional relationship between one and the other of the specific second partial regions are relatively different.
2. The image processing apparatus according to claim 1, wherein the image processing apparatus executes processing. the control unit specifies the second extended region corresponding to the first extended region based on a first feature point of the first extended region, a second feature point of the second extended region, and a predetermined feature point matching method using machine learning.
3. The image processing apparatus according to claim 1, wherein the image processing apparatus executes processing. the control unit generates the first extended region in which the first predetermined region is set to a size that avoids interference with the calculation of the first feature amount by the other of the specific first partial regions, and generates the second extended region in which the second predetermined region is set to a size that avoids interference with the calculation of the second feature amount by the other of the specific second partial regions.
3. The image processing apparatus according to claim 1, wherein the image processing apparatus executes processing. The first predetermined area and the second predetermined area are both blank spaces.
3. The image processing device according to claim 1, wherein the image processing device is a computer.