JP5888043B2 - Program, information processing method and information processing apparatus - Google Patents

Description

  The present invention relates to a program, an information processing method, and an information processing apparatus for detecting a change area from a difference image between two images.

  In recent years, there is a technique for collating two aerial images at different shooting points. For example, when performing alignment so that the same subject is in the same location on two images, the feature points are obtained in each of the two images, and the positions of the feature points on the new image and the old image are aligned. Done.

  On the other hand, the difference obtained after alignment between two images is due to the difference caused by the change of the subject shape and the difference in appearance caused by the difference in weather, season and angle. The resulting difference is included. “Difference in appearance” includes the position and intensity of the shadow of the building itself, the position and intensity of the shadow due to clouds, etc., and the difference in the appearance color between wet and dry conditions. .

  Therefore, by excluding the pixel value difference that occurs due to the difference in the appearance of the two images, the difference due to the change in the subject shape between the two images can be obtained appropriately.

  Focusing on the fact that the size and position of the shadow changes depending on the shooting date and time, there is a technique that prevents this shadow portion from being extracted as a difference.

JP-A-2005-173128 Japanese Patent Laid-Open No. 10-269347

  However, a huge amount of calculation is required to calculate each shadow region for an image having many buildings such as an urban area. In addition to the shadow, there is a factor that causes a difference in appearance, so the difference due to the change in the subject shape cannot be detected properly. In other words, in the prior art, when detecting a difference due to a change in subject shape between two images, many differences due to differences in appearance between images are detected, and many erroneous detections are included. There was a problem.

  Therefore, the disclosed technique aims to provide a program, an information processing method, and an information processing apparatus capable of reducing erroneous detection when detecting a difference due to a change in a subject shape between two images. .

A program according to an aspect of the disclosed technology acquires a first image and a second image that is different in shooting time from the first image, and sets a pair for each feature point of the first image. The second image feature point is extracted from a predetermined range including the position of the second image corresponding to the position of the feature point of the first image, and the first image and the second image Generating a differential image with the image, generating a polygon by specifying a portion having a difference equal to or greater than a predetermined threshold from the differential image, and out of the generated polygons other than the polygons including the extracted feature points of the pair The computer is caused to execute processing for detecting an area based on the polygon.

  According to the disclosed technique, it is possible to reduce false detection when detecting a difference due to a change in the subject shape between two images.

The figure which shows an example of the change area | region detection system in an Example. The block diagram which shows an example of the hardware constitutions of the information processing apparatus in an Example. The block diagram which shows an example of a functional structure of the information processing apparatus in an Example. The figure which shows an example of a 1st image. The figure which shows an example of a 2nd image. The figure which shows an example of the 1st image after color covering correction | amendment. The figure which shows an example of the 1st image after color simple correction | amendment. The figure which shows an example of the 2nd image after color covering correction | amendment. The figure which shows an example of the 2nd image after color simple correction | amendment. The figure which shows the 1st image before and after color normalization. The figure which shows the 2nd image before and after color normalization. The figure which shows an example of a template matching result. The figure which shows an example of a difference image. The figure which shows an example of the movement vector with respect to an expansion area | region. The figure which shows an example of the area | region detected and the area | region deleted. The figure which shows an example of a detection result. The flowchart which shows an example of the change area | region detection process in an Example. The flowchart which shows an example of the feature point pair extraction process in an Example.

  First, in the embodiment described below, a change point of a fixed asset is specified by comparing images including fixed assets taken from a satellite or an aircraft and extracting a difference. As an image to be compared, an aerial image or a satellite image may be used. The fixed assets are tangible fixed assets such as land and houses, and can be suitably applied particularly to houses.

  Therefore, an embodiment that reduces false detection when detecting a change location of a fixed asset by comparing two images will be described with reference to the accompanying drawings.

[Example]
<System>
FIG. 1 is a diagram illustrating an example of a changed area detection system 1 in the embodiment. The change area system 1 shown in FIG. 1 can be implemented using a cloud computing system. In the change area detection system 1 shown in FIG. 1, an information processing apparatus 10 as a server and information processing apparatuses 20, 30, and 40 as clients are connected via a network. The number of servers and clients is not limited to the example shown in FIG.

  In this change area detection system 1, the following two types of cloud service usage can be considered. First, the client side acquires satellite images, aerial images, etc., and sends the images to the server side, so that the server side extracts the change areas such as fixed assets from the satellite images, aerial images, and extracts them. This service notifies clients of results.

  The second is a service that holds satellite images, aerial images, and the like on the server side, extracts change areas such as fixed assets from the two images designated by the client, and notifies the client of the extraction results.

  The information processing apparatus 10 performs a process of extracting a change area such as a fixed asset from at least two images. The information processing apparatuses 20 to 40 use a change area detection service such as a fixed asset. Next, the information processing apparatus 10 will be described.

<Hardware>
FIG. 2 is a block diagram illustrating an example of a hardware configuration of the information processing apparatus 10 according to the embodiment. 2 includes at least a control unit 101, a main storage unit 102, an auxiliary storage unit 103, a communication unit 104, and a recording medium I / F unit 105. Each unit is connected via a bus so that data can be transmitted / received to / from each other.

  The control unit 101 is a CPU (Central Processing Unit) that performs control of each device, calculation of data, and processing in a computer. The control unit 101 is an arithmetic device that executes a program stored in the main storage unit 102 or the auxiliary storage unit 103. The control unit 101 receives data from the communication unit 104 or each storage unit, calculates, processes, and outputs the data. And output to each storage unit.

  Further, the control unit 101 performs a change area detection function by executing a change area detection program stored in the auxiliary storage unit 103, for example.

  The main storage unit 102 is a ROM (Read Only Memory), a RAM (Random Access Memory), or the like, and a storage device that stores or temporarily stores programs and data such as an OS and application software that are basic software executed by the control unit 101. It is.

  The auxiliary storage unit 103 is an HDD (Hard Disk Drive) or the like, and is a storage device that stores data related to application software or the like. Further, the auxiliary storage unit 103 may store a change area detection program acquired from the recording medium 106 or the like, a plurality of satellite images, or aerial images.

  The communication unit 104 performs wired or wireless communication. For example, the communication unit 104 acquires images from the information processing apparatuses 20 to 40 and receives service use instructions.

  A recording medium I / F (interface) unit 105 is an interface between the information processing apparatus 10 and a recording medium 106 (for example, a flash memory) connected via a data transmission path such as a USB (Universal Serial Bus).

  A predetermined program is stored in the recording medium 106, and the program stored in the recording medium 106 is installed in the information processing apparatus 10 via the recording medium I / F unit 105. The installed predetermined program can be executed by the information processing apparatus 10.

  If the recording medium 106 is an SD card, for example, the recording medium I / F unit 105 is an SD card slot.

<Function>
FIG. 3 is a block diagram illustrating an example of a functional configuration of the information processing apparatus 10 according to the embodiment. In the example illustrated in FIG. 3, the information processing apparatus 10 includes an image acquisition unit 200, a registration unit 201, a color normalization unit 202, a difference image generation unit 203, a polygon generation unit 204, and a feature point pair extraction unit 205. And a detection unit 206.

  The image acquisition unit 200 acquires a first image and a second image that is different in shooting time from the first image. The image acquisition unit 200 may acquire an image stored in the auxiliary storage unit 103, or may acquire an image transmitted from the information processing apparatus 10 or the like via the communication unit 104.

  The image acquisition unit 200 preferably acquires images taken at the same position using the same camera. The image acquisition unit 200 acquires, for example, an image obtained by photographing fixed assets such as a house and land from the sky, such as a satellite image and an aerial image. The image acquisition unit 200 outputs the acquired first image and second image to the alignment unit 201, the color normalization unit 202, and the feature point pair extraction unit 205.

  For example, the alignment unit 201 cuts out the center of the second image at% specified in the ini file, and performs template matching with the first image. Which of the first image and the second image is used as a reference can be changed as appropriate. The alignment unit 201 notifies the difference image generation unit 203 of the aligned information.

  In addition, the alignment unit 201 may perform alignment for a color-normalized first image and a second image, which will be described later. In addition, the alignment unit 201 does not need to perform alignment if two images are taken by the same camera and the position and angle are not changed.

  The color normalization unit 202 includes, for example, a color fog correction unit 221 that performs color fog correction and a color simple correction unit 222 that performs color simple correction in order to reduce differences due to differences in shooting time and shooting time. Color fog correction is a technique for correcting an image that has become entirely red due to, for example, sunset.

  The color fog correction unit 221 performs stretching so that 0.3% of each of the RGB histograms is included in 0,255. For example, when the image is covered with red, the color fog correction unit 221 extends the upper 0.3% of GB because it is about 230. The color fog correction unit 221 may perform known color fog correction in addition to the above.

  For example, after calculating each RGB histogram, the simple color correction unit 222 calculates an optimum Y value from color balance correction and RGB average, and performs LUT (look-up table) conversion. The color covering correction and the simple color correction are processes for performing color normalization in order to suppress a color difference between images. Note that the color covering correction and the simple color correction are not necessarily required in such a case because images taken at the same time, the same weather, and the like have a small color difference between the images.

  Further, the color normalization unit 202 may perform processing by appropriately combining histogram averaging (flattening), blurring, median expansion, color covering correction, simple color correction, and the like for color normalization.

  The difference image generation unit 203 generates a difference image between the first image and the second image acquired from the color normalization unit 202. For example, the difference image generation unit 203 calculates the distance between the pixels of the images aligned by the alignment unit 201 to generate a difference image. The difference image generation unit 203 outputs the generated difference image to the polygon generation unit 204.

  The polygon generation unit 204 generates a polygon from the difference image acquired from the difference image generation unit 203. For example, the polygon generation unit 204 generates a polygon by labeling a portion where the difference value of the difference image is equal to or greater than a predetermined threshold. The polygon generation unit 204 outputs the generated polygon to the detection unit 206.

  The feature point pair extraction unit 205 extracts feature points for the two images acquired from the color normalization unit 202 or for the two images acquired from the image acquisition unit 200 if color normalization is not performed. The feature point pair extraction unit 205 sets the feature point of the second image that is paired with each feature point of the first image to the second image corresponding to the position of the feature point of the first image. It extracts from within the predetermined range including the position. A method of extracting feature point pairs will be described later. The feature point pair extraction unit 205 outputs the extracted feature point pairs to the detection unit 206.

  The detection unit 206 detects an area based on a polygon other than the polygon including the feature point to be paired extracted by the feature point pair extraction unit 205 among the polygons generated by the polygon generation unit 204.

  The detection unit 206 includes, for example, a pair area deletion unit 261 and a shape determination unit 262. The pair area deletion unit 261 deletes the polygon including the feature point of the pair extracted by the feature point pair extraction unit 205 from the detection target. This is because the fact that there is a pair of feature points in a close region is considered to indicate that the feature points are slightly shifted. Therefore, the region with the pair of feature points is not a fixed asset change. I reckon.

  A region without a feature point pair indicates that there are no similar feature points in two images in a close region, and this region is a region where the fixed asset may be changed.

  The shape determining unit 262 determines whether the area of the polygon is equal to or smaller than a predetermined lower limit (first threshold) and is equal to or larger than a predetermined upper limit (second threshold). If the area of the polygon does not fall within the range from the lower limit value to the upper limit value, the shape determination unit 262 deletes it from the detection target. This is to prevent the detection of yes in a suitable area as a fixed asset.

  The shape determination unit 262 determines whether the shape of the polygon is a predetermined shape. If the polygon corresponds to a predetermined shape, the shape determination unit 262 deletes the polygon from the detection target. The shape determination unit 262 deletes the polygon from the detection target when the filling rate is smaller than the third threshold and the square rate is smaller than the fourth threshold.

  When the filling rate is smaller than the third threshold, the polygon may be a hollow rectangle, and when the square rate is smaller than the fourth threshold, the polygon becomes a key shape or L time shape. These shapes may not be suitable as fixed assets.

  Therefore, the shape determining unit 262 determines whether or not the polygon is appropriate as the shape of the fixed asset, and deletes the inappropriate polygon from the detection target.

  The detection unit 206 detects, among the generated polygons, the remaining polygons that are not deleted as an area where the fixed asset may be changed. 3 can be realized by, for example, the control unit 101 and the main storage unit 102 as a work memory shown in FIG.

  By having the above-described configuration, obviously, by deleting an area that is not a difference due to a fixed asset change from the detection target, it is possible to reduce false detections when detecting a fixed asset change area.

<Specific processing example of each part>
Next, a specific processing example of each unit will be described. In the example described below, a case where a region where a house has been changed (renovation, extension, new construction) is detected is taken as an example. Further, in the example described below, the following two images that are different at the time of shooting are used.

  FIG. 4 is a diagram illustrating an example of the first image. In the example illustrated in FIG. 4, the first image is an image in which an urban area is taken from the sky. The first image is an image taken on May 31, 2010. The first image is an image that is slightly covered with blue.

  FIG. 5 is a diagram illustrating an example of the second image. In the example illustrated in FIG. 5, the second image is an image that is substantially the same position as the first image and is different in shooting time. The second image is an image taken on March 31, 2007. The second image is an image that is slightly covered with green.

  The first image and the second image are preferably taken with the same camera, but may be different images as long as they are taken at the same position. Assume that the image acquisition unit 200 acquires the image shown in FIG. 4 and the image shown in FIG.

《Color normalization processing》
Next, a description will be given of a case where color covering correction and simple color correction are performed in order to suppress the color difference between images.

(First image)
The color fog correction unit 221 performs color fog correction on the first image. FIG. 6 is a diagram illustrating an example of a first image after color covering correction. In the example shown in FIG. 6, the first image after the color covering correction is slightly dark overall, but the blue covering of the area ar101 disappears.

  The simple color correction unit 222 performs simple color correction on the first image after color covering correction. FIG. 7 is a diagram illustrating an example of the first image after simple color correction. In the example shown in FIG. 7, the first image after the simple color correction is a little brighter than the image darkened by the color covering correction.

(Second image)
The color fog correction unit 221 performs color fog correction on the second image. FIG. 8 is a diagram illustrating an example of a second image after color covering correction. In the example shown in FIG. 8, the overall green covering disappears in the second image after the color covering correction.

  The simple color correction unit 222 performs simple color correction on the second image after color covering correction. FIG. 9 is a diagram illustrating an example of the second image after the simple color correction. In the example shown in FIG. 9, the second image after the simple color correction is slightly brighter than the image darkened by the color covering correction.

  FIG. 10 is a diagram illustrating a first image before and after color normalization. FIG. 10A shows a first image before color normalization, and FIG. 10B shows a first image after color normalization. Compared to the first image before color normalization shown in FIG. 10A, the first image after color normalization shown in FIG. .

  FIG. 11 is a diagram illustrating a second image before and after color normalization. FIG. 11A shows a second image before color normalization, and FIG. 11B shows a second image after color normalization. Compared with the second image before color normalization shown in FIG. 11A, the second image after color normalization shown in FIG. .

《Positioning process》
The alignment unit 201 performs template matching on the first image and the second image using RGB. The first image and the second image may be normalized images.

  FIG. 12 is a diagram illustrating an example of a template matching result. As shown in FIG. 12, even in two images taken of the same region, if the time of shooting is different, a small error due to parallax or the like occurs. Therefore, two images with different shooting times do not always match exactly even if they are the same object.

<< Differential image generation process >>
The difference image generation unit 203 generates a difference image between the first image and the second image that have been subjected to color normalization. The difference image is generated between the images after alignment. The difference image generation unit 203 calculates a distance between corresponding pixels in the two images and generates a difference image.

  For example, the difference image generation unit 203 generates a difference image between any one of RGB, YCbCr, CbCr only, YCbCr reduced Y by a predetermined amount, or only RG. The difference image generation unit 203 preferably performs RGB color difference (YCbCr) conversion and measures the distance for each band. At this time, the distance between Y is reduced to, for example, 1/4.

  As a result, it is possible to suppress the influence of the shadow and the brightness mismatch, and to ignore the change in brightness to some extent and to distinguish the color difference with priority.

  FIG. 13 is a diagram illustrating an example of the difference image. In the example illustrated in FIG. 13, the difference image generation unit 203 performs YCbCr conversion, and calculates a difference by setting Y to ¼. As shown in FIG. 13, many differences are extracted.

  In the example illustrated in FIG. 13, the area ar151 is an area that has been modified, the area ar152 is an area that has not been modified, but a difference is detected due to a difference in appearance or the like, and the area ar153 is This is an area where a difference is detected because the colors are different.

  Therefore, in the conventional technique, even if it is attempted to detect a change in fixed assets, a small error such as the area ar152 or the area ar253 is detected. However, in this embodiment, feature point pairs are extracted as shown below in order to reduce as much as possible errors caused by differences in appearance between images (region ar152, etc.).

《Feature point pair extraction processing》
Next, processing for extracting feature point pairs will be described. In the first place, since the two images are taken in the same area, if the fixed assets are not changed, the feature points should be paired within a predetermined range. Therefore, it is estimated that the region where a pair of feature points can be taken within a predetermined range is not a difference due to a change in fixed assets but a difference due to a difference in the appearance of the image.

  The feature point pair extraction unit 205 extracts feature points from each of the two images. For example, the feature point pair extraction unit 205 extracts 500 SURF (Speeded Up Robust Features) feature values for each image.

  In addition, the feature point pair extraction unit 205 holds information related to the extracted feature points. The information related to the feature point includes an edge having the feature point as one end point, edge information including the direction of the edge, and color information indicating a color around the feature point. In the color information, when an area is divided by an edge and there are a plurality of colors around the feature point, information on which color is located at which position with respect to the feature point is included in the color information.

The feature point pair extraction unit 205 extracts a feature point pair that satisfies a predetermined condition for each feature point extracted from the two images. The predetermined conditions are as follows.
(1) The feature points are within a predetermined range,
(2) Feature points are similar Condition (1) is a condition based on the idea that the same feature points should exist in the near range if they are the same object. When the feature point pair extraction unit 205 searches for the feature point of the second image corresponding to the feature point of the first image, the feature point pair extraction unit 205 sets the pair from the feature point of the first image within 30 pixels in both the x direction and the y direction. Search for feature points.

  Condition (2) is a condition based on the idea that the feature points should be similar if they are the same object. The feature point pair extraction unit 205 compares the information related to the feature points of the first image with the information related to the feature points of the second image for the feature points satisfying the condition (1), Determine whether to pair. As a result of the comparison, the feature point pair extraction unit 205 determines that the feature points are similar if the sum of the absolute difference values of each information is equal to or less than a threshold (fifth threshold), and performs pairing of the feature points. Do. As a target of similarity determination, a square error or the like may be used.

  As described above, the feature point pair extraction unit 205 can extract a feature point pair that satisfies the conditions (1) and (2). Hereinafter, a line segment connecting the paired feature points with each other is referred to as a movement vector.

  Assuming that there are no significant differences in the x and y directions for the positions in the two images for the same object, it is conceivable to obtain a movement vector within a predetermined range with respect to the x and y planes. The predetermined range is, for example, within 30 pixels from the target feature point in both the x and y directions. The feature point pair extraction unit 205 outputs the obtained feature point pair to the detection unit 206.

<Polygon generation processing>
The polygon generation unit 204 performs vectorization by labeling on the difference image generated by the difference image generation unit 203. For example, the polygon generation unit 204 performs labeling on a black portion of a binarized image. Accordingly, the polygon generation unit 204 can generate polygon data for each labeled region. A known technique may be applied to the generation of the polygon.

<< Detection process >>
Next, the fixed asset change area detection process will be described. First, the detection unit 206 deletes, from the detection target, a polygon that is estimated to have occurred without changing the fixed asset with respect to the polygon generated by the polygon generation unit 204.

  The pair area deletion unit 261 deletes the polygon including the feature point of the pair extracted by the feature point pair extraction unit 205 from the detection target. Further, the pair area deletion unit 261 may delete a polygon including the extracted feature points of the pair in a predetermined value or more from the detection target.

  Moreover, the pair area deletion unit 261 may delete the extracted pair based on the movement vector generated from the extracted feature point of the pair. This is because if the movement vector is not in the same direction and size as the surrounding movement vectors, there is a possibility that the pair has been erroneously paired.

  FIG. 14 is a diagram illustrating an example of a movement vector for an enlarged region. In the example shown in FIG. 14, feature points for this image are represented by dots, and this point and feature points of another paired image are connected by a line to represent a movement vector. That is, the tip of the line segment indicates the position of the feature point of another paired image.

  It has been experimentally found that the region having the movement vector shown in FIG. 14 is a region where the fixed asset has not been changed. Therefore, the pair area deletion unit 261 uses this experimental result to delete the area including the feature point pair from the detection target.

  On the other hand, there are cases where the movement vectors intersect, as in an area ar201 shown in FIG. In addition, as in the area ar202 illustrated in FIG. Further, as in a region ar203 illustrated in FIG. 14, a certain movement vector may face a different direction from other movement vectors.

  The movement vectors included in the areas ar201 to 203 shown in FIG. 14 may be movement vectors generated as a result of erroneous pairing. Therefore, the pair area deletion unit 261 deletes the pair of feature points forming the predetermined movement vector if the predetermined movement vector is not similar to the average vector of the movement vectors within the predetermined range from the movement vector. To do.

  The pair area deletion unit 261 obtains a value (SAD) obtained by adding the absolute difference values of the movement vectors in the x and y directions for determining the similarity of the movement vectors, and if this value is within a predetermined threshold, the similarity is determined. It can be judged.

  Thereby, the reliability of the polygon produced | generated by the difference resulting from the difference in an image can be improved.

  Further, the shape determination unit 262 determines a polygon that is not appropriate for the shape of a house or the like based on the generated area and shape of the polygon, and deletes the polygon from the detection target. For example, the shape determining unit 262 deletes, from the detection target, polygons whose polygon area does not fall within a predetermined range from the lower limit value to the upper limit value, or whose squareness ratio or filling rate is lower than a predetermined threshold value.

  The detection unit 206 outputs a region including polygons that remain without being deleted from the detection target as detection results.

  FIG. 15 is a diagram illustrating an example of a detected area and a deleted area. In the example illustrated in FIG. 15, with respect to the difference image, the area ar301 indicates a detected area, the area ar302 indicates an area where the polygon area is deleted without satisfying the condition, and the area ar303 indicates a filling rate. Indicates a region that is deleted because the square is lower than the threshold, a region ar304 indicates a region that is deleted because the square ratio is lower than the threshold, and a region ar305 indicates a region that is deleted based on the movement vector.

  “Deleting based on a movement vector” means, for example, deleting an area including both or one of feature points forming a movement vector. As described above, the pair region deletion unit 261 may delete the movement vector itself if the movement vector is not similar to the surrounding average vector.

  According to the example shown in FIG. 15, the number of areas ar305 is about three times the number of areas ar301. By deleting this area ar305 from the detection target, erroneous detection can be reduced. The area ar301 includes areas other than the areas where the fixed assets are changed, but the detection unit 206 can reduce false detections by deleting the areas ar302 to ar305.

<Detection result>
Next, a result detected by the information processing apparatus 10 will be described. FIG. 16 is a diagram illustrating an example of a detection result. In the example shown in FIG. 16, the area ar301 shown in FIG. 15 is applied to the second image. The detection unit 206 may apply the detection area to the first image.

  An area ar401 illustrated in FIG. 16 is an area where there is a difference between the first image and the second image, that is, an area where the house has been renovated. A region ar401 illustrated in FIG. 16 is appropriately detected as a region ar301 illustrated in FIG.

  An area ar402 illustrated in FIG. 16 indicates an area appropriately deleted based on the movement vector. The area ar402 is an area where there is no renovation of the house, and it can be seen that the difference caused by the difference in the appearance of the image can be appropriately deleted.

  With the above configuration, the information processing apparatus 10 can reduce erroneous detection when detecting a difference due to a change in the subject shape between two images.

<Operation>
Next, the operation of the information processing apparatus 10 will be described. FIG. 17 is a flowchart illustrating an example of the change area detection process in the embodiment.

  In step S101, the image acquisition unit 200 inputs two images. Here, the first image is a reference image, and the second image is a comparative image.

  In step S102, the alignment unit 201 aligns the position of the comparison image with the reference image. The alignment unit 201 performs template matching in a predetermined area at the center of the image, for example.

  In step S103, the color normalization unit 202 normalizes the colors of the two images. Thereby, the color difference between images can be suppressed.

  In step S104, the feature point pair extraction unit 205 extracts a feature point pair that satisfies a predetermined condition between the two images. This process will be described later with reference to FIG. By this processing, it becomes possible to extract an area where there is no change in a fixed asset (for example, a house).

  In step S105, the difference image generation unit 203 generates a difference image from the two images. Thereby, the area including the area where the fixed asset is changed can be detected.

  In step S106, the polygon generation unit 204 generates a polygon by labeling the difference image.

  In step S107, the pair area deletion unit 261 deletes the area including the extracted pair feature points from the detection target. Thereby, the erroneous detection part can be deleted.

  In step S108, the shape determining unit 262 deletes the polygon determined based on the polygon area, the filling rate, and the square rate from the detection target. Thereby, it is possible to delete the elongated shape, the key shape, the donut shape, and the polygon whose area is outside the predetermined range.

  In step S109, the detection unit 206 makes it possible to identify a region including a polygon remaining as a detection target by, for example, surrounding the reference image with a line, and detect a reference image that can identify the detected region. Output as.

  Steps S102 and S103 are not necessarily required processes. Further, the order of processing is not limited to the order shown in FIG. 17, and step S102 may be performed before step S105, and steps S103 and S104 may be performed before step S107.

  Next, detailed processing in step S104 will be described. FIG. 18 is a flowchart illustrating an example of feature point pair extraction processing in the embodiment.

  In step S201, the feature point pair extraction unit 205 selects one feature point of the first image (reference image).

  In step S202, the feature point pair extraction unit 205 sets a range within a predetermined range from the position corresponding to the selected feature point for the second image (comparison image).

  In step S203, the feature point pair extraction unit 205 extracts a feature point that is most similar to the selected feature point of the first image from among the feature points that are within the set predetermined range. The feature point pair extraction unit 205 can determine similarity by comparing information related to feature points (feature point information).

  In step S204, the feature point pair extraction unit 205 determines whether the most similar feature point satisfies the condition (2) described above. If the condition (2) is satisfied (step S204—YES), the process proceeds to step S205, and if the condition (2) is not satisfied (step S204—NO), the process proceeds to step S206.

  In step S205, the feature point pair extraction unit 205 sets feature points satisfying the condition (2) as a pair.

  In step S206, the feature point pair extraction unit 205 determines whether there is an unprocessed feature point in the first image. If there is no unprocessed feature point (step S206—YES), this process is terminated, and the process proceeds to step S105. If there is an unprocessed feature point (step S206—NO), the process returns to step S201, and other feature points are found. Selected.

  As described above, according to the disclosed embodiment, it is possible to reduce erroneous detection when detecting a difference due to a change in the subject shape between two images.

  Further, according to the disclosed embodiment, it is possible to conduct business using image data taken by a satellite or an aircraft. For example, by comparing two images taken by different satellites or aircraft at the time of shooting and extracting the difference, it is possible to specify the land or building that has been renovated or renovated.

  Here, according to the disclosed technology, for local government employees who perform management of fixed assets and calculation of fixed asset tax, it is necessary to identify a place to be observed based on two images of areas to be managed. Therefore, the disclosed technique is considered useful. In addition, what is necessary is just to set an appropriate predetermined value for each threshold value mentioned above by experiment etc. FIG.

[Modification]
Note that by recording a program for realizing the processing described in each of the above-described embodiments on a recording medium, the processing in each of the embodiments can be performed by a computer. For example, it is possible to record the program on a recording medium and cause the computer to read the recording medium on which the program is recorded, thereby realizing the processing described above.

  The recording medium is a recording medium that records information optically, electrically, or magnetically, such as a CD-ROM, flexible disk, magneto-optical disk, etc., and information is electrically recorded such as ROM, flash memory, etc. Various types of recording media such as a semiconductor memory can be used. This recording medium does not include a carrier wave.

  Although the embodiments have been described in detail above, the invention is not limited to the specific embodiments, and various modifications and changes can be made within the scope described in the claims.

In addition, the following additional notes are disclosed regarding each of the above embodiments.
(Appendix 1)
Obtaining a first image and a second image different from the first image at the time of shooting;
A feature point of the second image paired with each feature point of the first image is within a predetermined range including the position of the second image corresponding to the position of the feature point of the first image. Extracted from
Generating a difference image between the first image and the second image;
Generate a polygon from the difference image,
A program for causing a computer to execute processing for detecting a region based on a polygon other than the polygon including the extracted feature point of the extracted pair among the generated polygons.
(Appendix 2)
The extraction process is:
The program according to claim 1, wherein information relating to the feature point of the first image and information relating to the feature point of the second image are compared to determine whether to make a pair.
(Appendix 3)
The program according to claim 2, wherein the information on the feature point includes edge information of an edge having the feature point as one end point, and color information around the feature point.
(Appendix 4)
The process to detect is
The program according to any one of supplementary notes 1 to 3, wherein a polygon to be deleted from a detection target is determined based on an area and a shape of the polygon.
(Appendix 5)
The process of generating the difference image includes
The program according to any one of appendices 1 to 4, wherein color difference conversion is performed on the first image and the second image, and the difference image is generated between images whose luminance is reduced by a predetermined amount.
(Appendix 6)
The process to detect is
The program according to any one of appendices 1 to 5, wherein a pair of vectors determined to be different from surrounding vectors is deleted based on a vector connecting feature points of the extracted pairs.
(Appendix 7)
Obtaining a first image and a second image different from the first image at the time of shooting;
A feature point of the second image paired with each feature point of the first image is within a predetermined range including the position of the second image corresponding to the position of the feature point of the first image. Extracted from
Generating a difference image between the first image and the second image;
Generate a polygon from the difference image,
An information processing method in which a computer executes a process of detecting a region based on a polygon other than the polygon including the extracted feature point of the extracted pair among the generated polygons.
(Appendix 8)
An acquisition unit that acquires a first image and a second image that is different from the first image at the time of shooting;
A feature point of the second image paired with each feature point of the first image is within a predetermined range including the position of the second image corresponding to the position of the feature point of the first image. An extraction unit for extracting from,
A difference image generation unit for generating a difference image between the first image and the second image;
A polygon generation unit for generating a polygon from the difference image;
An information processing apparatus comprising: a detection unit that detects an area based on a polygon other than the polygon including the extracted feature point of the extracted pair among the generated polygons.
(Appendix 9)
A color normalization unit that performs color normalization on the first image and the second image;
The extraction unit and the difference image generation unit are:
The information processing apparatus according to appendix 8, which performs processing on the color-normalized first image and second image.

DESCRIPTION OF SYMBOLS 1 Change area | region detection system 10 Information processing apparatus 101 Control part 102 Main memory part 103 Auxiliary memory part 200 Image acquisition part 201 Positioning part 202 Color normalization part 203 Difference image generation part 204 Polygon generation part 205 Feature point pair extraction part 206 Detection 261 Pair region deletion unit 262 Shape determination unit

Claims (8)

Obtaining a first image and a second image different from the first image at the time of shooting;
A feature point of the second image paired with each feature point of the first image is within a predetermined range including the position of the second image corresponding to the position of the feature point of the first image. Extracted from
Generating a difference image between the first image and the second image;
Generate a polygon by identifying a portion where the difference is greater than or equal to a predetermined threshold from the difference image,
A program for causing a computer to execute processing for detecting a region based on a polygon other than the polygon including the extracted feature point of the extracted pair among the generated polygons. The extraction process is:
The program according to claim 1, wherein information relating to the feature point of the first image and information relating to the feature point of the second image are compared to determine whether to make a pair.   3. The program according to claim 2, wherein the information on the feature point includes edge information of an edge having the feature point as one end point and color information around the feature point. The process to detect is
The program according to any one of claims 1 to 3, wherein a polygon to be deleted from a detection target is determined based on an area and a shape of the polygon. The process of generating the difference image includes
The program according to any one of claims 1 to 4, wherein color difference conversion is performed on the first image and the second image, and the difference image is generated between images whose luminance is reduced by a predetermined amount. The process to detect is
The program according to any one of claims 1 to 5, wherein a pair of vectors determined to be different from surrounding vectors is deleted based on a vector connecting feature points among the extracted pairs. Obtaining a first image and a second image different from the first image at the time of shooting;
A feature point of the second image paired with each feature point of the first image is within a predetermined range including the position of the second image corresponding to the position of the feature point of the first image. Extracted from
Generating a difference image between the first image and the second image;
Generate a polygon by identifying a portion where the difference is greater than or equal to a predetermined threshold from the difference image,
An information processing method in which a computer executes a process of detecting a region based on a polygon other than the polygon including the extracted feature point of the extracted pair among the generated polygons. An acquisition unit that acquires a first image and a second image that is different from the first image at the time of shooting;
A feature point of the second image paired with each feature point of the first image is within a predetermined range including the position of the second image corresponding to the position of the feature point of the first image. An extraction unit for extracting from,
A difference image generation unit for generating a difference image between the first image and the second image;
A polygon generation unit for generating a polygon by specifying a portion having a difference equal to or greater than a predetermined threshold from the difference image;
An information processing apparatus comprising: a detection unit that detects an area based on a polygon other than the polygon including the extracted feature point of the extracted pair among the generated polygons.