Bright spot detection device and bright spot detection method

Abstract

Even when multiple peak points are close to each other, it is possible to individually determine whether each peak point is the target bright spot. [Solution] The bright spot detection device 10 includes a peak point detection unit 11 that detects N local peak points (N is an integer of 2 or more) included in the input image; a mask image generation unit 12 that generates a mask image including N mask regions in which a predetermined range centered on each detected peak point is masked; an antimask image generation unit 13 that generates N different antimask images in which the mask is removed in one of the N mask regions, and when superimposed on the input image, each of the N peak points is included individually in the superimposed input image; and a determination unit 14 that determines, based on predetermined conditions, whether the peak points included individually in each superimposed input image obtained by superimposing the N different antimask images on the input image are target bright spots.

Description

Technical Field

[0001] The present disclosure relates to a bright point detection device and a bright point detection method.

Background Art

[0002] Conventionally, there has been known a bright point detection technique for detecting a particularly bright and shining portion (hereinafter referred to as a "bright point") in an image or video and obtaining important information from the detected bright point. The bright point detection technique is used in various fields such as, for example, astronomy, surveillance systems, medical image processing, industrial robots, autonomous driving, and environmental monitoring.

[0003] In relation to the bright point detection technique, for example, Patent Document 1 discloses an example of a method for detecting peak points. In this method, first, a digital image signal is input to an input terminal, and a maximum value indicating the maximum luminance value in one screen of the input digital image signal and its coordinate position (peak point position) are detected by a maximum value detection unit. Data indicating the detected maximum value and peak point position is supplied to an image masking unit. The image masking unit sets the luminance of pixel data within a preset screen range centered on the peak point position detected by the maximum value detection unit to the minimum value (i.e., masks) among the image information for one screen in the above-described digital image signal, and outputs a digital image signal in which the pixel data in other screen ranges has the original luminance value. The output digital image signal is further input to another maximum value detection unit, and by this maximum value detection unit, a maximum value and a peak point position are detected from the digital image signal, and data indicating the detected maximum value and peak point position is supplied to another image masking unit. Hereinafter, in this method, the detection of the maximum value and the peak point position, the masking process, and the output of a new digital image signal are repeated, and the peak point position is detected for the number of repetitions.

[0004] Incidentally, when detecting the peak point position using the method disclosed in Patent Document 1, the peak point at that position may contain noise such as radiation in addition to the bright spot that is the intended to be detected, such as a star. Therefore, as a method for determining whether this peak point is the bright spot that is the intended to be detected (hereinafter referred to as the "target bright spot") or noise, a method is sometimes employed that evaluates the correlation between the luminance distribution at this peak point and a two-dimensional Gaussian distribution. This method utilizes the fact that, for example, if the target bright spot is a star, the luminance distribution in an image containing that bright spot takes the form of a Gaussian distribution centered on that bright spot, but the luminance distribution in an image containing noise does not take such a Gaussian distribution. [Prior art documents] [Patent Documents]

[0005] [Patent Document 1] Japanese Patent Application Publication No. 2-014387 [Overview of the project] [Problems that the invention aims to solve]

[0006] However, in the above determination method, when multiple peak points are close to each other, if we try to evaluate the correlation between each of these close peak points and the Gaussian distribution, the luminance distribution of one of the close peak points and the luminance distribution of the other peak points will influence each other, and there is a risk that what should be determined as multiple target bright spots will be determined as a single target bright spot.

[0007] This disclosure was made to solve the above-mentioned problems and aims to provide a bright spot detection device that can individually determine whether each peak point is a target bright spot, even when multiple peak points are close to each other. [Means for solving the problem]

[0008] The bright spot detection device according to this disclosure comprises: a peak point detection unit that detects N local peak points (N is an integer of 2 or more) contained in an input image based on image information representing an input image; a mask image generation unit that generates a mask image containing N mask regions by masking a predetermined range centered on each detected peak point of the image information; an antimask image generation unit that generates N antimask images by demasking one of the N mask regions in the mask image, wherein the mask regions to be demasked differ from one another, so that when superimposed on the input image, each of the N peak points is included individually in the superimposed input image; and a determination unit that determines, based on predetermined conditions, whether or not a peak point included individually in each of the superimposed input images obtained by sequentially superimposing the N antimask images onto the input image is a bright spot to be detected. [Effects of the Invention]

[0009] According to this disclosure, with the configuration described above, even when multiple peak points are close to each other, it becomes possible to individually determine whether or not each peak point is the target bright spot. [Brief explanation of the drawing]

[0010] [Figure 1] This figure shows an example configuration of the bright spot detection device according to Embodiment 1. [Figure 2] Figures 2A and 2B are flowcharts showing examples of operation of the bright spot detection device according to Embodiment 1. [Figure 3] Figure 3A shows an example of an input image in Embodiment 1, Figure 3B shows an example of a peak point in Embodiment 1, and Figure 3C shows an example of a mask image in Embodiment 1. [Figure 4]Figure 4A shows an example of extracting a peripheral image from the input image in Embodiment 1 that corresponds to the portion where the mask area overlaps; Figure 4B shows an example of extracting an image from the mask image in Embodiment 1 that corresponds to the portion in Figure 4A; Figure 4C shows an example of an antimask image in Embodiment 1; and Figure 4D shows an example of an input image with an antimask image superimposed in Embodiment 1. [Figure 5] Figure 5A shows an example of extracting a peripheral image from the input image in Embodiment 1 that corresponds to the portion where the mask area overlaps; Figure 5B shows an example of extracting an image from the mask image in Embodiment 1 that corresponds to the portion in Figure 5A; Figure 5C shows an example of an antimask image in Embodiment 1; and Figure 5D shows an example of an input image with an antimask image superimposed in Embodiment 1. [Figure 6] Figures 6A and 6B show examples of the hardware configuration of the bright spot detection device according to Embodiment 1. [Figure 7] Figures 7A and 7B are diagrams illustrating an example of the processing of the determination unit in Embodiment 2. [Figure 8] This figure shows an example configuration of a bright spot detection device according to Embodiment 3. [Figure 9] This flowchart shows an example of the operation of the bright spot detection device according to Embodiment 3. [Figure 10] This diagram illustrates an example of the processing in the complementary processing unit in Embodiment 3. [Figure 11] This figure shows an example configuration of a bright spot detection device according to Embodiment 4. [Figure 12] This flowchart shows an example of the operation of the bright spot detection device according to Embodiment 4. [Figure 13] Figures 13A, 13B, and 13C illustrate an example of the processing of the peak point exclusion section in Embodiment 4. [Figure 14] This flowchart shows an example of the operation of the bright spot detection device according to Embodiment 5. [Figure 15]FIG. 15A, FIG. 15B, and FIG. 15C are diagrams for explaining an example of the process of the peak point exclusion unit in Embodiment 5. [Figure 16] It is a flowchart showing an operation example of the bright point detection device according to Embodiment 6. [Figure 17] FIG. 17A and FIG. 17B are diagrams for explaining an example of the process of the peak point exclusion unit in Embodiment 6.

Embodiments for Carrying Out the Invention

[0011] Hereinafter, the embodiments will be described in detail with reference to the drawings. Embodiment 1. FIG. 1 is a block diagram showing a configuration example of the bright point detection device 10 according to Embodiment 1. As shown in FIG. 1 for example, the bright point detection device 10 includes a peak point detection unit 11, a mask image generation unit 12, an anti-mask image generation unit 13, and a determination unit 14.

[0012] The peak point detection unit 11 acquires an input image that is the object of bright point detection. When the peak point detection unit 11 acquires the input image, based on the image information indicating the acquired input image, it detects local peak points included in the input image. For example, the peak point detection unit 11 detects N local peak points included in the input image based on the image information indicating the input image. N is an integer of 2 or more. Note that the peak point detection unit 11 can detect local peak points included in the input image using a known method.

[0013] The peak point detection unit 11 outputs information indicating the detected peak points (hereinafter referred to as "peak point information") and the image information indicating the input image to the mask image generation unit 12. The peak point information also includes information indicating the positions of the peak points in the input image.

[0014] The mask image generation unit 12 acquires peak point information and image information representing the input image from the peak point detection unit 11. Once the mask image generation unit 12 acquires the peak point information and image information representing the input image from the peak point detection unit 11, it generates a mask image that includes N mask regions by masking a predetermined range centered on each peak point of the image information representing the input image based on the peak point information.

[0015] The mask is, for example, a circle with a predetermined radius. The mask image is, for example, a binary image in which the pixel values ​​of the masked area are "0" and the pixel values ​​of the other areas are "1". The mask image generation unit 12 outputs information indicating the generated mask image (hereinafter referred to as "mask image information") to the antimask image generation unit 13.

[0016] The antimask image generation unit 13 obtains mask image information from the mask image generation unit 12. Upon obtaining the mask image information from the mask image generation unit 12, the antimask image generation unit 13 generates N different antimask images by demasking one of the N mask regions in the mask image indicated by the mask image information. The N different antimask images are such that, because the mask regions that are demasked are different from each other, when superimposed on the input image, each of the N peak points is included individually in the superimposed input image.

[0017] The processing performed by the antimask image generation unit 13 differs depending on whether or not there are overlapping areas among the N mask regions in the mask image. To accommodate this case distinction, the antimask image generation unit 13 is configured to include a distance calculation unit 131 and an antimask processing unit 132.

[0018] The distance calculation unit 131 calculates the distance between each of the N detected peak points. The distance calculation unit 131 outputs information indicating the calculated distance to the antimask processing unit 132.

[0019] The anti-mask processing unit 132 determines, based on the distance between peak points and the radius of the circular mask, whether or not there is a portion where multiple mask regions overlap among the N mask regions in the mask image.

[0020] For example, the antimask image generation unit 13 compares the distance between peak points with twice the radius of the circular mask, and determines that there is no overlap in the mask regions if the distances between peak points are greater than or equal to twice the radius of the circular mask. On the other hand, the antimask processing unit 132 compares the distance between peak points with twice the radius of the circular mask, and determines that there is an overlap in the mask regions if either of the distances between peak points is less than twice the radius of the circular mask.

[0021] The antimask processing unit 132 generates N different antimask images by demasking one of the N mask regions in the mask image if there is no overlap between multiple mask regions in the mask image. As described above, the N different antimask images are such that, when superimposed on the input image, each of the N peak points is included individually in the superimposed input image, because the mask regions that are demasked are different from each other.

[0022] On the other hand, if there is an overlapping portion of multiple mask regions among the N mask regions in the mask image, the antimask processing unit 132 generates N different antimask images by demasking one of the N mask regions in the mask image. In this case, the antimask processing unit 132 demasks the portion of the mask region that does not overlap with the other mask regions in the overlapping portion.

[0023] The antimask image generation unit 13 outputs N different pieces of information (hereinafter referred to as "antimask image information") indicating the generated antimask image to the determination unit 14.

[0024] The determination unit 14 acquires N antimask image information from the antimask image generation unit 13. Once the determination unit 14 has acquired N antimask image information from the antimask image generation unit 13, it sequentially superimposes the N different antimask images indicated by the acquired antimask image information onto the input image. For each of the resulting superimposed input images, the determination unit 14 determines, based on predetermined conditions, whether or not the peak point contained in the input image alone is a bright spot to be detected (target bright spot).

[0025] For example, if the bright spot to be detected is a star, the determination unit 14 evaluates the correlation between the brightness distribution of the peak point included alone in the input image with the antimask image superimposed and the two-dimensional Gaussian distribution, and determines whether or not the peak point is a star based on the evaluation result. Specifically, the determination unit 14 calculates a correlation coefficient between the brightness distribution of the peak point included alone in the input image with the antimask image superimposed and the two-dimensional Gaussian distribution as the correlation, and compares the calculated correlation coefficient with a predetermined threshold. If the calculated correlation coefficient is greater than or equal to the threshold, the determination unit 14 determines that the peak point is a star. On the other hand, if the calculated correlation coefficient is less than the threshold, the determination unit 14 determines that the peak point is noise. The determination unit 14 outputs information indicating the result of the determination to the outside.

[0026] Next, we will describe an example of the operation of the bright spot detection device 10 shown in Figure 1. Figures 2A and 2B are flowcharts illustrating an example of the operation of the bright spot detection device 10.

[0027] First, the peak point detection unit 11 detects local peak points contained in the input image based on image information representing the input image (step ST1). For example, the peak point detection unit 11 detects N local peak points contained in the input image based on image information representing the input image, where N is an integer of 2 or more. The peak point detection unit 11 can detect local peak points contained in the input image using known methods.

[0028] Next, the mask image generation unit 12 generates a mask image containing N mask regions by masking a predetermined range centered on each peak point detected in step ST1 with respect to the image information representing the input image (step ST2). The mask is, for example, a circle with a predetermined radius. The mask image is also a binary image in which the pixel values ​​of the masked regions are "0" and the pixel values ​​of the other regions are "1".

[0029] An example of an input image and a mask image is shown in Figure 3. For example, when there is an input image as shown in Figure 3A, the peak point detection unit 11 detects local peak points contained in the input image based on the image information representing the input image. Here, the peak point detection unit 11 detects three peak points P1 to P3 as shown in Figure 3B. The mask image generation unit 12 generates a mask image that includes three mask regions as shown in Figure 3C by masking a predetermined range centered on each of the detected peak points.

[0030] Next, the antimask image generation unit 13 generates N different antimask images, each of which is obtained by demasking one of the N mask regions in the mask image. These N antimask images are such that, when superimposed on the input image, each of the N peak points is included individually in the superimposed input image, because the mask regions to be demasked are different from each other. (Step ST3) Demasking means changing the pixel values ​​of a predetermined range of regions centered on the peak points from "0" to "1".

[0031] The processing in step ST3 consists of substeps ST31 and ST32, as shown in Figure 2B, for example. Specifically, in substep ST31, the distance calculation unit 131 calculates the distance between each of the N peak points detected in step ST1 (substep ST31). The distance calculation unit 131 outputs data indicating the calculated distance to the antimask processing unit 132.

[0032] The antimask processing unit 132 generates N different antimask images, each of which is obtained by demasking one of the N mask regions in the mask image. These N antimask images are such that, because the demasked regions are different from each other, when superimposed on the input image, each of the N peak points is included individually in the superimposed input image (substep ST32). At this time, the antimask processing unit 132 determines whether or not there is an overlap between multiple mask regions among the N mask regions, based on the distance between the peak points calculated in substep ST31 and the radius of the circular mask.

[0033] For example, the antimask image generation unit 13 compares the distance between peak points with twice the radius of the circular mask, and determines that there are no overlapping areas between multiple mask regions if the distances between peak points are greater than or equal to twice the radius of the circular mask. In this case, the antimask processing unit 132 generates N different antimask images, each of which is obtained by demasking one of the N mask regions in the mask image. Because the mask regions to be demasked are different from each other, when superimposed on the input image, each of the N peak points is included individually in the superimposed input image.

[0034] For example, when there are three masked areas in the mask image, the antimask processing unit 132 generates three different antimask images. In each of the three antimask images, the masked areas that are unmasked are different. When each of the three antimask images is superimposed on the input image, each of the three peak points will be included individually in the superimposed input image. For example, when the first antimask image is superimposed on the input image, the superimposed input image will include the first peak point individually. When the second antimask image is superimposed on the input image, the superimposed input image will include the second peak point individually. When the third antimask image is superimposed on the input image, the superimposed input image will include the third peak point individually.

[0035] On the other hand, the anti-masking processing unit 132 compares the distance between the peak points with twice the radius of the circular mask, and determines that there is an overlapping area between multiple mask regions if either of the distances between the peak points is less than twice the radius of the circular mask. For example, in the example shown in Figure 3B, the distance between peak point P2 and peak point P3 is less than twice the radius of the circular mask, and the two mask regions centered on these two peak points overlap as shown in Figure 3C.

[0036] In this case, the antimask processing unit 132 extracts the surrounding image from the input image corresponding to the portion where the mask area overlaps, as shown in Figure 4A, for example. The antimask processing unit 132 also extracts the image from the mask image corresponding to the portion in Figure 4A, i.e., the surrounding image of the portion where the mask area overlaps, as shown in Figure 4B, for example.

[0037] Next, the antimask processing unit 132 generates two antimask images by removing the mask in one of the two mask regions from the peripheral image extracted from the mask image, and by removing the mask in each of the two mask regions, the two antimask images are such that when superimposed on the input image, each of the two peak points is included individually in the superimposed input image, as the mask regions to be removed are different.

[0038] For example, the antimask processing unit 132 generates two types of antimask images as shown in Figure 4C. In the two types of antimask images, the masked areas that are unmasked are different. In this case, when the antimask processing unit 132 unmasks one masked area, it does not unmask the parts that overlap with the other masked area. That is, it leaves the pixel values ​​as "0". In other words, when the antimask processing unit 132 unmasks one masked area, it only unmasks the parts that do not overlap with the other masked area.

[0039] When each of the two antimask images is superimposed on the input image, as shown in Figure 4D, for example, each of the two peak points will be included individually in the superimposed input image. In other words, in the input image on which the antimask image is superimposed, one of the two peak points is masked, and the other is included individually. As a result, the determination unit 14, described later, can determine whether the individually included peak point is the target bright spot without being affected by the brightness distribution of other peak points.

[0040] In the example above, we explained the case where two mask regions centered on two peak points overlap, but the same applies to the case where three mask regions centered on three peak points overlap.

[0041] In this case, the antimask processing unit 132 extracts the surrounding image from the input image corresponding to the portion where the mask area overlaps, as shown in Figure 5A, for example. The antimask processing unit 132 also extracts the image from the mask image corresponding to the portion in Figure 5A, i.e., the surrounding image of the portion where the mask area overlaps, as shown in Figure 5B, for example.

[0042] Next, the antimask processing unit 132 generates three different antimask images by removing the mask in one of the three mask regions from the peripheral image extracted from the mask image. These three antimask images are generated such that, when superimposed on the input image, each of the three peak points is included individually in the superimposed input image, because the mask regions to be removed are different from each other.

[0043] For example, the antimask processing unit 132 generates three different antimask images as shown in Figure 5C. In each of the three antimask images, the masked areas that are unmasked are different. In this case, when the antimask processing unit 132 unmasks one masked area, it does not unmask the parts that overlap with other masked areas. That is, it leaves the pixel values ​​as "0". In other words, when the antimask processing unit 132 unmasks one masked area, it only unmasks the parts that do not overlap with other masked areas.

[0044] When each of the three antimask images is superimposed on the input image, as shown in Figure 5D, for example, each of the three peak points will be included individually in the superimposed input image. In other words, in the input image on which the antimask image is superimposed, two of the three peak points are masked, and one is included individually. As a result, the determination unit 14, described later, can determine whether the individually included peak point is the target bright spot without being affected by the brightness distribution of other peak points.

[0045] The above explanation described the case where mask regions centered on two or three peak points overlap, but the same applies to the case where mask regions centered on four or more peak points overlap.

[0046] Furthermore, the above explanation described an example in which the antimask image generation unit 13 generates an antimask image even when there is no overlap between multiple mask regions among the N mask regions. However, the antimask image generation unit 13 may omit the generation of an antimask image if, for example, there is no overlap between multiple mask regions in the mask image and the N mask regions are far apart from each other; in other words, if the N peak points are far apart from each other. This is because, if the peak points are far apart from each other, it may be possible to determine whether a certain peak point is the target bright spot without superimposing an antimask image, without being affected by the brightness distribution of other peak points. In this case, the distance between the N peak points required to omit the generation of an antimask image can be appropriately set according to the design.

[0047] Next, the determination unit 14 sequentially superimposes the N antimask images generated in step ST3 onto the input image to obtain each of the superimposed input images, and determines, based on predetermined conditions, whether or not the peak point included in the input image alone is a bright spot to be detected (step ST4).

[0048] For example, if the bright spot to be detected is a star, the determination unit 14 evaluates the correlation between the brightness distribution of the peak point included alone in the input image with the superimposed antimask image and the two-dimensional Gaussian distribution, and determines whether or not the peak point is a star based on the evaluation result. Specifically, the determination unit 14 calculates the correlation coefficient between the brightness distribution of the peak point included alone in the input image with the superimposed antimask image and the two-dimensional Gaussian distribution, and compares the calculated correlation coefficient with a predetermined threshold. If the calculated correlation coefficient is greater than or equal to the threshold, the determination unit 14 determines that the peak point is a star. On the other hand, if the calculated correlation coefficient is less than the threshold, the determination unit 14 determines that the peak point is not a star (i.e., it is noise). In this case, since the input image with the superimposed antimask image includes the peak point alone, the determination unit 14 can accurately determine whether or not a given peak point is a star without being affected by the brightness distribution of the other peak point, even if there is another peak point in the vicinity of that peak point.

[0049] As described above, the bright spot detection device 10 comprises a peak point detection unit 11, a mask image generation unit 12, an anti-mask image generation unit 13, and a determination unit 14. This allows the bright spot detection device 10 to individually determine whether each peak point is a target bright spot, even when multiple peak points detected in the input image are close to each other, without being affected by the brightness distribution of other nearby peak points.

[0050] Next, with reference to Figure 6, an example of the hardware configuration of the bright spot detection device 10 according to Embodiment 1 will be described. The functions of the peak point detection unit 11, the mask image generation unit 12, the antimask image generation unit 13, and the determination unit 14 in the bright spot detection device 10 are realized by a processing circuit. The processing circuit may be dedicated hardware as shown in Figure 6A, or it may be a CPU (Central Processing Unit, central processing unit, processing unit, arithmetic unit, microprocessor, microcomputer, processor, or DSP (Digital Signal Processor)) 52 that executes a program stored in memory 53, as shown in Figure 6B.

[0051] If the processing circuit is dedicated hardware, the processing circuit 51 may be, for example, a single circuit, a composite circuit, a programmed processor, a parallel programmed processor, an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or a combination thereof. The functions of each part, the peak point detection unit 11, the mask image generation unit 12, the antimask image generation unit 13, and the determination unit 14, may be implemented individually by the processing circuit 51, or the functions of each part may be implemented together by the processing circuit 51.

[0052] When the processing circuit is a CPU 52, the functions of the peak point detection unit 11, the mask image generation unit 12, the antimask image generation unit 13, and the determination unit 14 are realized by software, firmware, or a combination of software and firmware. The software and firmware are written as programs and stored in memory 53. The processing circuit realizes the functions of each unit by reading and executing the programs stored in memory 53. In other words, the bright spot detection device 10 has a memory for storing programs that, when executed by the processing circuit, result in the execution of each step shown in Figure 2, for example. These programs can also be said to cause the computer to execute the procedures and methods of the peak point detection unit 11, the mask image generation unit 12, the antimask image generation unit 13, and the determination unit 14. Here, memory 53 includes, for example, non-volatile or volatile semiconductor memory such as RAM (Random Access Memory), ROM (Read Only Memory), flash memory, EPROM (Erasable Programmable ROM), EEPROM (Electrically EPROM), magnetic disks, flexible disks, optical disks, compact disks, minidiscs, or DVDs (Digital Versatile Discs).

[0053] Furthermore, the functions of the peak point detection unit 11, mask image generation unit 12, antimask image generation unit 13, and determination unit 14 may be partially implemented by dedicated hardware and partially implemented by software or firmware. For example, the peak point detection unit 11 can be implemented by a processing circuit as dedicated hardware, while the mask image generation unit 12, antimask image generation unit 13, and determination unit 14 can be implemented by a processing circuit reading and executing a program stored in memory 53.

[0054] Thus, the processing circuit can realize each of the above-mentioned functions through hardware, software, firmware, or a combination thereof.

[0055] As described above, according to this embodiment 1, the bright spot detection device 10 includes: a peak point detection unit 11 that detects N local peak points (N is an integer of 2 or more) included in an input image based on image information indicating the input image; a mask image generation unit 12 that generates a mask image including N mask regions by masking a predetermined range centered on each detected peak point with respect to the image information; an antimask image generation unit 13 that generates N antimask images in which the mask is released in one of the N mask regions in the mask image, and by making the mask regions to be released different from each other, such that when superimposed on the input image, each of the N peak points is included individually in the superimposed input image; and a determination unit 14 that determines, based on predetermined conditions, whether or not a peak point included individually in each of the superimposed input images obtained by sequentially superimposing the N antimask images on the input image is a bright spot to be detected. As a result, the bright spot detection device 10 according to Embodiment 1 can individually determine whether each peak point is a target bright spot, even when multiple peak points are close to each other.

[0056] Furthermore, the antimask image generation unit 13 includes an antimask processing unit 132 that, in the mask image, if there is a portion where multiple mask regions overlap among the N mask regions, removes the mask from one mask region that is located in the overlapping portion, for the portion that does not overlap with the other mask regions located in the overlapping portion. As a result, the bright spot detection device 10 according to Embodiment 1 can accurately generate N different antimask images.

[0057] Furthermore, the mask image generation unit 12 defines each of the N mask regions as a circular region having the same radius, the antimask image generation unit 13 includes a distance calculation unit 131 that calculates the distance between each of the N peak points detected in the input image, and the antimask processing unit 132 determines whether or not there is a portion in the mask image where multiple mask regions overlap among the N mask regions based on the distance calculated by the distance calculation unit 131 and the radius of the mask region. As a result, the bright spot detection device 10 according to Embodiment 1 can accurately determine whether or not there is a portion in the mask image where multiple mask regions overlap among the N mask regions.

[0058] Furthermore, the determination unit 14 evaluates the correlation between the luminance distribution of a peak point included alone in the input image on which the antimask image is superimposed and the two-dimensional Gaussian distribution, and determines whether or not the peak point is a star based on the result of this evaluation. As a result, the bright spot detection device 10 according to Embodiment 1 can accurately determine whether or not a peak point is a star based on the correlation between the luminance distribution of the peak point and the two-dimensional Gaussian distribution.

[0059] Furthermore, the determination unit 14 calculates a correlation coefficient between the luminance distribution of a peak point included alone in the input image on which the antimask image is superimposed, and a two-dimensional Gaussian distribution. If the calculated correlation coefficient is greater than or equal to a threshold, the determination unit 14 determines that the peak point is a star; if the calculated correlation coefficient is less than the threshold, the determination unit 14 determines that the peak point is noise. As a result, the bright spot detection device 10 according to Embodiment 1 can accurately determine whether or not a peak point is a star based on the correlation coefficient between the luminance distribution of the peak point and a two-dimensional Gaussian distribution.

[0060] Embodiment 2. Embodiment 1 describes a bright spot detection device 10 that can individually determine whether each peak point is a target bright spot, even when multiple peak points are close to each other. Embodiment 2 describes a bright spot detection device 10b that simplifies the determination process by the determination unit 14 compared to the bright spot detection device 10 according to Embodiment 1.

[0061] The configuration example of the bright spot detection device 10b according to Embodiment 2 is the same as the configuration example of the bright spot detection device 10 according to Embodiment 1 shown in Figure 1. However, in Embodiment 2, the processing content by the determination unit 14 differs from that of Embodiment 1.

[0062] In Embodiment 2, the determination unit 14 sequentially superimposes N different antimask images onto the input image to obtain each of the superimposed input images. Based on predetermined conditions, the determination unit 14 determines whether or not a peak point included in the input image alone is a bright spot to be detected.

[0063] Generally, if the detected bright spot is a star, the determination unit 14 in Embodiment 2 evaluates the correlation between the brightness distribution of the peak point and the two-dimensional Gaussian distribution indicated by the code GD, as shown in Figure 7A, for example, and determines whether or not the peak point is a star based on the evaluation result. Specifically, the determination unit 14 in Embodiment 2 calculates the correlation coefficient between the brightness distribution of the peak point and the two-dimensional Gaussian distribution, and compares the calculated correlation coefficient with a threshold. If the correlation coefficient is greater than or equal to the threshold, the determination unit 14 determines that the peak point is a star, while if the correlation coefficient is less than the threshold, it determines that the peak point is noise.

[0064] Here, in Embodiment 1, the determination unit 14 calculates a correlation coefficient between the luminance distribution of a peak point and a two-dimensional Gaussian distribution, even for a peak point that is included alone in the input image on which the antimask image is superimposed, as shown in Figure 7B, for example. The determination unit 14 compares the calculated correlation coefficient with a threshold, and if the correlation coefficient is greater than or equal to the threshold, it determines that the peak point is a star, and if the correlation coefficient is less than the threshold, it determines that the peak point is noise.

[0065] The determination unit 14 in Embodiment 2 also performs determination in basically the same manner as described above, but in this case, when calculating the correlation coefficient, the determination unit 14 in Embodiment 2 considers the luminance value in the mask region as an invalid value and excludes it from the calculation of the correlation coefficient.

[0066] For example, the determination unit 14 in Embodiment 2 can calculate the difference between an ideal Gaussian distribution and the brightness distribution of the image centered on the peak point, and calculate a correlation coefficient (correlation value) by dividing the sum of the difference values ​​per pixel by the number of pixels. In this case, the determination unit 14 in Embodiment 2 calculates the correlation coefficient after excluding the brightness values ​​in the masked area from both the difference calculation and the count as pixels. This allows the determination unit 14 to simplify the processing when calculating the correlation coefficient and to improve the accuracy of the calculated correlation coefficient.

[0067] As described above, according to this second embodiment, when the determination unit 14 calculates the correlation coefficient, it considers the luminance value in the mask region of the input image on which the antimask image is superimposed as an invalid value and excludes it from the calculation of the correlation coefficient. As a result, in addition to the effects of the first embodiment, the bright spot detection device 10b can simplify the process of calculating the correlation coefficient between the luminance distribution of the peak point and the two-dimensional Gaussian distribution compared to the bright spot detection device 10 according to the first embodiment, and can calculate the correlation coefficient with high accuracy.

[0068] Embodiment 3. Embodiment 1 describes a bright spot detection device 10 that can individually determine whether each peak point is a target bright spot, even when multiple peak points are close to each other. Embodiment 3 describes a bright spot detection device 10c that can reduce the amount of processing required for determination by the determination unit 14 compared to the bright spot detection device 10 according to Embodiment 1.

[0069] Figure 8 shows an example of the configuration of the bright spot detection device 10c according to Embodiment 3. The bright spot detection device 10c according to Embodiment 3 has an additional interpolation processing unit 15 compared to the bright spot detection device 10 according to Embodiment 1 shown in Figure 1. The other components of the bright spot detection device 10c according to Embodiment 3 are the same as those of the bright spot detection device 10 according to Embodiment 1 shown in Figure 1, and therefore the same reference numerals are used and their descriptions are omitted.

[0070] The interpolation processing unit 15 synthesizes an input image, which has an antimask image superimposed on it and contains a single peak point, with an interpolation image. The interpolation image is generated by the interpolation processing unit 15. For example, the interpolation processing unit 15 generates the interpolation image by rotating the input image with the antimask image superimposed by a predetermined angle, or by flipping the input image with the antimask image superimposed vertically, horizontally, or vertically and horizontally.

[0071] The interpolation processing unit 15 interpolates the pixel values ​​of the mask region in the input image superimposed with the antimask image by combining the generated interpolation image with the input image superimposed with the antimask image, thereby interpolating the pixel values ​​of the mask region in the interpolation image corresponding to that mask region. At this time, the interpolation processing unit 15 may, for example, invalidate the pixel values ​​of the mask region in the interpolation image beforehand so that the pixel values ​​of the mask region in the interpolation image are not reflected in the input image superimposed with the antimask image.

[0072] The interpolation processing unit 15 may perform the above synthesis in such a manner that it fills in the pixel values ​​of the mask region included in the input image on which the antimask image is superimposed with the pixel values ​​of the region of the interpolation image corresponding to the mask region. Alternatively, the interpolation processing unit 15 may set the brightness value of the mask region included in the input image on which the antimask image is superimposed to "0", and then perform the above synthesis in such a manner that the region with the brightness value set to "0" and the region of the interpolation image corresponding to the mask region ultimately remain.

[0073] The determination unit 14 then performs the above-described determination on the image obtained by the interpolation processing unit 15. The determination method by the determination unit 14 is the same as the method described in Embodiment 1. For example, the determination unit 14 calculates a correlation coefficient between the brightness distribution of the peak point included in the image obtained by the interpolation and a two-dimensional Gaussian distribution as a correlation. If the calculated correlation coefficient is greater than or equal to a threshold, the determination unit 14 determines that the peak point is a star. If the calculated correlation coefficient is less than the threshold, the determination unit 14 determines that the peak point is noise.

[0074] Figure 9 is a flowchart showing an example of operation of the bright spot detection device 10c according to Embodiment 3. In the flowchart shown in Figure 9, step ST3-2 is added between step ST3 and step ST4 compared to the flowchart showing an example of operation of the bright spot detection device 10 according to Embodiment 1 described in Figure 2. Note that each process in steps ST1 to ST3 in Figure 9 is the same as each process in steps ST1 to ST3 shown in Figure 2, so a further explanation is omitted.

[0075] In step ST3-2, the interpolation processing unit 15 combines the input image, which has an antimask image superimposed on it and contains a single peak point, with the interpolation image. The interpolation processing unit 15 generates the interpolation image by rotating the input image with the antimask image superimposed by a predetermined angle, or by flipping the input image with the antimask image superimposed vertically or horizontally. Then, by performing the above synthesis, the interpolation processing unit 15 interpolates the pixel values ​​of the mask region included in the input image with the antimask image superimposed with the pixel values ​​of the region of the interpolation image corresponding to the mask region.

[0076] An example of the processing in this case will be explained with reference to Figure 10. In Figure 10, the leftmost image is an input image with an antimask image superimposed on it, containing only a single peak point. Here, a predetermined range has been extracted from the input image. The central image in Figure 10 is the interpolation image generated by the interpolation processing unit 15. Here, the interpolation processing unit 15 generates the interpolation image by flipping the input image with the antimask image superimposed on it, shown on the leftmost part of Figure 10, both vertically and horizontally.

[0077] The interpolation processing unit 15 combines these input images and interpolation images to generate an image in which, as shown on the far right of Figure 10, the pixel values ​​of the mask region in the input image are interpolated with the pixel values ​​of the region in the interpolation image corresponding to the mask region, and the pixel values ​​of the mask region in the interpolation image are interpolated with the pixel values ​​of the region in the input image corresponding to the mask region. At this time, the interpolation processing unit 15 may, for example, invalidate the pixel values ​​of the mask region included in the interpolation image beforehand before synthesis, so that the pixel values ​​of the mask region included in the interpolation image are not reflected in the input image on which the antimask image is superimposed.

[0078] In step ST3-2, after the interpolation processing unit 15 has performed the above processing, in step ST4, the determination unit 14 makes the same determination as in Embodiment 1 described above regarding the image obtained by the interpolation processing unit 15.

[0079] In this way, the interpolation processing unit 15 substantially removes the mask region from the image that the determination unit 14 is trying to determine by performing the above processing, and the determination unit 14 then makes a determination on the image from which the mask region has been removed. As a result, the determination unit 14 can reduce the amount of processing required for determination compared to when the determination is made while considering the mask region, which is advantageous in terms of the amount of processing required for determination compared to Embodiment 1.

[0080] As described above, according to this embodiment 3, the bright spot detection device 10c includes an interpolation processing unit 15 that interpolates the pixel values ​​of the mask region included in the input image with the mask region using the pixel values ​​of the region of the interpolation image corresponding to the mask region. This interpolation processing unit 15 combines an input image with an antimask image superimposed on it, in which a peak point is included alone, and an interpolation image obtained by rotating the input image with the antimask image superimposed on it by a predetermined angle, or by flipping the input image with the antimask image superimposed on it vertically, horizontally, or vertically and horizontally. The determination unit 14 determines, based on predetermined conditions, whether a peak point included alone in the image obtained by interpolation by the interpolation processing unit 15 is a bright spot to be detected. As a result, the bright spot detection device 10c can reduce the amount of processing required for determination by the determination unit 14, in addition to the effects of embodiment 1, and is more advantageous than embodiment 1 in this respect.

[0081] Embodiment 4. Embodiment 1 described a bright spot detection device 10 that can individually determine whether each peak point is a target bright spot, even when multiple peak points are close to each other. Embodiment 4 describes a bright spot detection device 10d that is expected to improve the determination accuracy of the determination unit 14 and reduce the amount of processing required for determination, compared to the bright spot detection device 10 according to Embodiment 1.

[0082] Figure 11 shows an example of the configuration of the bright spot detection device 10d according to Embodiment 4. The bright spot detection device 10d according to Embodiment 4 has a peak point exclusion unit 16 added to the bright spot detection device 10 according to Embodiment 1 shown in Figure 1. The other components of the bright spot detection device 10d according to Embodiment 4 are the same as those of the bright spot detection device 10 according to Embodiment 1 shown in Figure 1, so the same reference numerals are used and their descriptions are omitted. Furthermore, the bright spot detection device 10d according to Embodiment 4 is assumed to be mainly mounted on a satellite, and the bright spots (target bright spots) that the bright spot detection device 10d detects are assumed to be stars.

[0083] The peak point exclusion unit 16 excludes peak points that meet predetermined conditions from among the N peak points detected by the peak point detection unit 11 from the image information representing the input image.

[0084] For example, the peak point exclusion unit 16 excludes from the image information representing the input image any peak points among the N peak points detected by the peak point detection unit 11 that do not show continuity in the time-series changes in position or brightness in a predetermined number of frames before and after the frame relating to the input image. The peak point exclusion unit 16 can acquire the predetermined number of frames before and after the frame relating to the input image from an imaging device (not shown) mounted on the satellite. Furthermore, the number of frames before and after the frame relating to the input image to be acquired by the peak point exclusion unit 16 can be arbitrarily set according to the design, etc.

[0085] The mask image generation unit 12, by masking a predetermined range centered on peak points that were not excluded by the peak point exclusion unit 16 with respect to the image information representing the input image, generates a mask image containing M mask regions (where M is an integer between 2 and N) instead of a mask image containing N mask regions.

[0086] The antimask image generation unit 13 generates M antimask images, which are obtained by demasking one of the M mask regions in the mask image, instead of N antimask images. By making the demasking regions different from each other, the M antimask images are generated such that when superimposed on the input image, each of the M peak points is included individually in the superimposed input image.

[0087] Figure 12 is a flowchart showing an example of operation of the bright spot detection device 10d according to Embodiment 4. In the flowchart shown in Figure 12, step ST1-2 is added between step ST1 and step ST2 compared to the flowchart showing an example of operation of the bright spot detection device 10 according to Embodiment 1 described in Figure 2. Note that the processes of steps ST1 and ST4 in Figure 12 are the same as the processes of steps ST1 and ST4 shown in Figure 2, so a further explanation is omitted.

[0088] In step ST1-2, the peak point exclusion unit 16 excludes from the image information representing the input image any peak points that meet predetermined conditions from among the N peak points detected by the peak point detection unit 11. For example, the peak point exclusion unit 16 excludes from the image information representing the input image any peak points from among the N peak points in the input image detected by the peak point detection unit 11 that do not show continuity in the temporal position or brightness changes in a predetermined number of frames before and after the frame relating to the input image.

[0089] An example of the processing in this case will be explained with reference to Figure 13. Figure 13A is a frame of the image that is one frame earlier in time than the frame of the input image, Figure 13B is the frame of the input image, i.e., the frame that is subject to the peak point exclusion process, and Figure 13C is a frame of the image that is one frame later in time than the frame of the input image.

[0090] In Figures 13A-13C, the star symbols indicate peak points that show continuous changes in position over time. Specifically, the four peak points P11-P14, all marked with stars, move approximately the same distance from right to left in the diagram as time progresses. Peak points that move at a nearly constant velocity over time in this manner are highly likely to be stars.

[0091] On the other hand, the circles indicate peak points that do not show continuity in their temporal positional changes. That is, the peak points P21 to P27, which are circles, appear suddenly in one frame and disappear in the next, showing that they move in a non-continuous manner over time. It is assumed that peak points that move in this manner are highly likely to be noise such as radiation or thermal noise, rather than stars. Therefore, the peak point exclusion unit 16 excludes the peak points P23, P24, and P25, which are circles that do not show continuity in their temporal positional changes, from the image information representing the input image.

[0092] In step ST2, the mask image generation unit 12 generates a mask image that includes M (in this case, 4) mask regions by masking a predetermined range centered on the peak points P11 to P14 of the star marks that were not excluded by the peak point exclusion unit 16, with respect to the image information representing the input image.

[0093] In step ST3, the antimask image generation unit 13 generates M different antimask images, each of which is obtained by demasking one of the M mask regions in the mask image. By making the demasking regions different from each other, the M antimask images are generated such that when superimposed on the input image, each of the M peak points is included individually in the superimposed input image.

[0094] In the example above, the peak point exclusion unit 16 described an example in which peak points that do not have continuity in their time-series position changes are excluded from the image information representing the input image. However, the peak point exclusion unit 16 may also exclude, for example, peak points that do not have continuity in their time-series brightness changes from the image information representing the input image.

[0095] In this way, the peak point exclusion unit 16 can efficiently separate non-stationary peak points such as radiation or thermal noise from the data to be judged by excluding peak points that meet predetermined conditions. As a result, in Embodiment 4, compared to Embodiment 1, an improvement in the judgment accuracy of the judgment unit 14 and a reduction in the processing load for judgment can be expected.

[0096] As described above, according to this embodiment 4, the bright spot detection device 10d includes a peak point exclusion unit 16 that excludes peak points that meet predetermined conditions from the N peak points detected by the peak point detection unit 11 from the image information representing the input image, the mask image generation unit 12 generates a mask image containing M mask regions (where M is an integer between 2 and N) instead of a mask image containing N mask regions by masking a predetermined range centered on the peak points that were not excluded by the peak point exclusion unit 16 with respect to the image information representing the input image, and the antimask image generation unit 13 generates M antimask images instead of N antimask images, in which the mask is released in one of the M mask regions in the mask image, and by making the mask regions to be released different from each other, the antimask image generates M antimask images such that when superimposed with the input image, each of the M peak points is included individually in the superimposed input image. As a result, in addition to the effects of Embodiment 1, the bright spot detection device 10d is expected to offer improved judgment accuracy by the judgment unit 14 and reduced processing load compared to Embodiment 1.

[0097] Furthermore, the peak point exclusion unit 16 excludes from the image information representing the input image any peak points among the N peak points in the input image detected by the peak point detection unit 11 that do not show continuity in the temporal position or brightness changes in a predetermined number of frames before and after the frame relating to the input image. As a result, the bright spot detection device 10d can efficiently separate non-stationary peak points such as radiation or thermal noise from the object to be judged.

[0098] Furthermore, the bright spot detection device 10d is mounted on the satellite, and the bright spots that the determination unit 14 uses for determination are stars. As a result, the bright spot detection device 10d is expected to improve the accuracy of the determination unit 14 when determining whether or not a peak point is a star, and to reduce the amount of processing required for determination.

[0099] Embodiment 5. In Embodiment 4, a bright spot detection device 10d was described, which is expected to improve the accuracy of the determination unit 14 and reduce the amount of processing required for determination, compared to the bright spot detection device 10 according to Embodiment 1. In Embodiment 5, a bright spot detection device 10e is described from a different perspective than Embodiment 4, which is expected to improve the accuracy of the determination unit 14 and reduce the amount of processing required for determination.

[0100] The configuration example of the bright spot detection device 10e according to Embodiment 5 is the same as the configuration example of the bright spot detection device 10d according to Embodiment 4 shown in Figure 11, but in Embodiment 5, the processing content by the peak point exclusion unit 16 differs from that of Embodiment 4. The bright spot detection device 10e according to Embodiment 5 is also assumed to be mainly mounted on a satellite, and the bright spots (target bright spots) detected by the bright spot detection device 10e are assumed to be stars.

[0101] In Embodiment 5, the peak point exclusion unit 16 compares multiple images with different fields of view in advance and records peak points whose positions do not change even when the fields of view are different. Then, the peak point exclusion unit 16 excludes from the image information representing the input image the peak points that correspond to the peak points recorded above, out of the N peak points in the input image detected by the peak point detection unit 11. The peak point exclusion unit 16 can acquire multiple images with different fields of view from an imaging device (not shown) mounted on the satellite.

[0102] The mask image generation unit 12, by masking a predetermined range centered on peak points that were not excluded by the peak point exclusion unit 16 with respect to the image information representing the input image, generates a mask image containing M mask regions (where M is an integer between 2 and N) instead of a mask image containing N mask regions.

[0103] The antimask image generation unit 13 generates M antimask images, which are obtained by demasking one of the M mask regions in the mask image, instead of N antimask images. By making the demasking regions different from each other, the M antimask images are generated such that when superimposed on the input image, each of the M peak points is included individually in the superimposed input image.

[0104] Figure 14 is a flowchart showing an example of operation of the bright spot detection device 10e according to Embodiment 5. In the flowchart shown in Figure 14, step ST1-2 is added between step ST1 and step ST2 compared to the flowchart showing an example of operation of the bright spot detection device 10 according to Embodiment 1 described in Figure 2. Note that the processes of steps ST1 and ST4 in Figure 14 are the same as the processes of steps ST1 and ST4 shown in Figure 2, so a further explanation is omitted.

[0105] In step ST1-2, the peak point exclusion unit 16 excludes peak points that meet predetermined conditions from among the N peak points detected by the peak point detection unit 11 from the image information representing the input image. For example, the peak point exclusion unit 16 compares multiple images with different shooting fields in advance and records peak points whose positions do not change even when the shooting fields are different. Then, the peak point exclusion unit 16 excludes from the image information representing the input image the peak points that correspond to the peak points recorded above from among the N peak points in the input image detected by the peak point detection unit 11.

[0106] An example of the processing in this case will be explained with reference to Figure 15. Figure 15A is a frame relating to an input image captured in a certain field of view, Figure 15B is a frame relating to an input image captured in a field of view different from that of Figure 15A, and Figure 15C is a frame relating to an input image captured in a field of view yet different from that of Figures 15A and 15B.

[0107] In Figures 15A to 15C, it can be seen that the peak points indicated by symbols P31 to P33 are located in the same position in the upper right corner of the image, even though the input images were taken in different fields of view. It is highly probable that such peak points are not stars, but rather defects caused by damage to the photosensitive element mounted on the imaging device. Therefore, the peak point exclusion unit 16 excludes peak points corresponding to those whose position does not change even when the field of view is different from the image information representing the input image.

[0108] In step ST2, the mask image generation unit 12 generates a mask image that includes M mask regions by masking a predetermined range centered on the peak points that were not excluded by the peak point exclusion unit 16 with respect to the image information representing the input image.

[0109] In step ST3, the antimask image generation unit 13 generates M different antimask images, each of which is obtained by demasking one of the M mask regions in the mask image. By making the demasking regions different from each other, the M antimask images are generated such that when superimposed on the input image, each of the M peak points is included individually in the superimposed input image.

[0110] Thus, in imaging devices mounted on satellites, damage to the photosensitive element can cause defective areas of the photosensitive element to be consistently recorded as peak points in the input image. The peak point exclusion unit 16 excludes peak points corresponding to such peak points caused by defects in the photosensitive element from the image information representing the input image in advance. As a result, in Embodiment 5, compared to Embodiment 1, an improvement in the judgment accuracy of the judgment unit 14 and a reduction in the processing load for judgment can be expected.

[0111] As described above, according to this embodiment 5, the peak point exclusion unit 16 compares multiple images with different shooting fields in advance, records peak points whose positions do not change even when the shooting fields are different, and excludes the peak points corresponding to the recorded peak points from the image information representing the input image, out of the N peak points in the input image detected by the peak point detection unit 11. As a result, in addition to the effects of embodiment 1, the bright spot detection device 10e can be expected to improve the judgment accuracy of the judgment unit 14 and reduce the amount of processing required for judgment compared to embodiment 1. Furthermore, the bright spot detection device 10e can efficiently separate peak points caused by defects due to damage to the photosensitive element from the object to be judged.

[0112] Embodiment 6. In Embodiment 1, a bright spot detection device 10d was described, which is expected to improve the accuracy of the determination unit 14 and reduce the amount of processing required for determination, compared to the bright spot detection device 10 according to Embodiment 1. In Embodiment 6, a bright spot detection device 10f is described from a different perspective than Embodiment 4, which is expected to improve the accuracy of the determination unit 14 and reduce the amount of processing required for determination.

[0113] The configuration example of the bright spot detection device 10f according to Embodiment 6 is the same as the configuration example of the bright spot detection device 10d according to Embodiment 4 shown in Figure 11, but in Embodiment 6, the processing content by the peak point exclusion unit 16 differs from that of Embodiment 4. The bright spot detection device 10f according to Embodiment 6 is also assumed to be mainly mounted on a satellite, and the bright spots (target bright spots) detected by the bright spot detection device 10f are assumed to be stars.

[0114] In Embodiment 6, the peak point exclusion unit 16 compares the input image with a star map in which known stars are pre-registered, and excludes from the image information representing the input image the peak points corresponding to stars registered in the star map, out of the N peak points in the input image detected by the peak point detection unit 11. The information representing the star map is pre-stored, for example, in a storage unit (not shown) provided in the bright spot detection device 10f.

[0115] The mask image generation unit 12, by masking a predetermined range centered on peak points that were not excluded by the peak point exclusion unit 16 with respect to the image information representing the input image, generates a mask image containing M mask regions (where M is an integer between 2 and N) instead of a mask image containing N mask regions.

[0116] The antimask image generation unit 13 generates M antimask images, which are obtained by demasking one of the M mask regions in the mask image, instead of N antimask images. By making the demasking regions different from each other, the M antimask images are generated such that when superimposed on the input image, each of the M peak points is included individually in the superimposed input image.

[0117] Figure 16 is a flowchart showing an example of operation of the bright spot detection device 10f according to Embodiment 6. In the flowchart shown in Figure 16, step ST1-2 is added between step ST1 and step ST2 compared to the flowchart showing an example of operation of the bright spot detection device 10 according to Embodiment 1 described in Figure 2. Note that the processes of steps ST1 and ST4 in Figure 16 are the same as the processes of steps ST1 and ST4 shown in Figure 2, so a further explanation is omitted.

[0118] In step ST1-2, the peak point exclusion unit 16 compares the input image with a star map in which known stars are pre-registered, and excludes from the image information representing the input image the peak points that correspond to stars registered in the star map, out of the N peak points in the input image detected by the peak point detection unit 11.

[0119] An example of the processing in this case will be explained with reference to Figure 17. Figure 17A is a star map in which known stars have been pre-registered, and Figure 17B is a frame related to the input image, i.e., a frame that will be subject to the peak point exclusion process.

[0120] The peak point exclusion unit 16 compares the star map shown in Figure 17A with the input image in the frame shown in Figure 17B. Here, the peak point exclusion unit 16 determines that the peak points of the star symbols that have corresponding (almost identical) positions and brightness in both images are peak points corresponding to stars. The peak point exclusion unit 16 then excludes the peak points that it has determined to be corresponding to stars from the image information representing the input image.

[0121] As a result, the image information representing the input image retains peak point P41, indicated by a circle, and peak point P42, indicated by a cross. These peak points are highly likely to be noise such as radiation or thermal noise, or planets or artificial objects other than stars registered in the star map.

[0122] Furthermore, the bright spot detection device 10f can determine whether peak point P41 is noise by evaluating the correlation between the luminance distribution of peak point P41 and the two-dimensional Gaussian distribution described above using the determination unit 14. In addition, the bright spot detection device 10f can determine whether peak point P42 is a planet or artificial object other than a star registered in the star map using known methods for identifying planets or artificial objects.

[0123] In step ST2, the mask image generation unit 12 generates a mask image that includes M (in this case, 2) mask regions by masking a predetermined range centered on peak points P41 to P42 that were not excluded by the peak point exclusion unit 16 with respect to the image information representing the input image.

[0124] In step ST3, the antimask image generation unit 13 generates M different antimask images, each of which is obtained by demasking one of the M mask regions in the mask image. By making the demasking regions different from each other, the M antimask images are generated such that when superimposed on the input image, each of the M peak points is included individually in the superimposed input image.

[0125] In this way, the peak point exclusion unit 16 excludes peak points that are determined to correspond to stars registered in the star map from the image information representing the input image by referring to the star map. As a result, in Embodiment 6, compared to Embodiment 1, an improvement in the accuracy of the determination unit 14 and a reduction in the amount of processing required for determination can be expected.

[0126] As described above, according to this embodiment 6, the peak point exclusion unit 16 compares the input image with a star map in which known stars are pre-registered, and excludes from the image information representing the input image the peak points corresponding to stars registered in the star map out of the N peak points in the input image detected by the peak point detection unit 11. As a result, in addition to the effects of embodiment 1, the bright spot detection device 10f can be expected to reduce the amount of processing required for determination by the determination unit 14 compared to embodiment 1, and the probability of detecting planets or artificial objects other than stars can be increased.

[0127] Although preferred embodiments have been described in detail above, the invention is not limited to the embodiments described above, and various modifications and substitutions can be made to the embodiments described above without departing from the scope of the claims.

[0128] Furthermore, this disclosure allows for free combination of each embodiment, modification of any component of each embodiment, or omission of any component in each embodiment.

[0129] The various aspects of this disclosure are summarized below as an appendix.

[0130] (Note 1) A peak point detection unit detects N local peak points (where N is an integer of 2 or more) contained in an input image based on image information representing the input image. A mask image generation unit generates a mask image that includes N mask regions by masking a predetermined range centered on each of the detected peak points of the image information. An antimask image generation unit generates N antimask images obtained by demasking one of the N mask regions in the aforementioned mask image, wherein the mask regions to be demasked differ from one another, so that when superimposed with the input image, each of the N peak points is included individually in the superimposed input image. A determination unit determines, based on predetermined conditions, whether a peak point included alone in each of the superimposed input images obtained by sequentially superimposing the N antimask images onto the input image is a bright spot to be detected. A bright spot detection device comprising the above components. (Note 2) The antimask image generation unit, In the aforementioned mask image, if there is a portion where multiple mask regions overlap among the N mask regions, the anti-masking processing unit includes a mechanism that removes the mask from one mask region located in the overlapping portion, specifically the portion that does not overlap with the other mask regions located in that overlapping portion. The bright spot detection device according to Appendix 1, characterized in that it is a bright spot detection device. (Note 3) The mask image generation unit sets each of the N mask regions as a circular region having the same radius. The antimask image generation unit, The input image includes a distance calculation unit that calculates the distance between each of the N peak points detected, The aforementioned anti-masking processing unit is Based on the distance calculated by the distance calculation unit and the radius of the mask region, it is determined whether or not there is an overlapping portion among the N mask regions in the mask image. The bright spot detection device according to Appendix 2, characterized by the features described above. (Note 4) The determination unit, The correlation between the luminance distribution of a peak point included alone in the input image overlaid with the aforementioned antimask image and a two-dimensional Gaussian distribution is evaluated, and based on the results of this evaluation, it is determined whether or not the peak point is a star. A bright spot detection device according to any one of the appendices 1 to 3, characterized by the above. (Note 5) The determination unit, As part of the correlation, the correlation coefficient is calculated between the luminance distribution of a peak point included alone in the input image on which the antimask image is superimposed, and a two-dimensional Gaussian distribution. If the calculated correlation coefficient is greater than or equal to a threshold, the peak point is determined to be a star; if the calculated correlation coefficient is less than the threshold, the peak point is determined to be noise. The bright spot detection device described in Appendix 4, characterized by the features described herein. (Note 6) The determination unit, When calculating the correlation coefficient, the luminance values ​​in the mask region of the input image on which the anti-mask image is superimposed are considered invalid values ​​and are excluded from the calculation of the correlation coefficient. The bright spot detection device according to Appendix 5, characterized by the features described herein. (Note 7) An input image superimposed with the antimask image, in which the aforementioned peak point is included alone, By rotating the input image on which the antimask image is superimposed by a predetermined angle, or by flipping the input image on which the antimask image is superimposed vertically, horizontally, or vertically and horizontally, and then compositing it with a complementary image obtained, The system includes an interpolation processing unit that interpolates the pixel values ​​of the mask region included in the input image on which the antimask image is superimposed with the pixel values ​​of the region of the interpolation image corresponding to the mask region. The determination unit, Based on the predetermined conditions, it is determined whether the peak point included alone in the image obtained by the interpolation process is a bright spot to be detected. A bright spot detection device according to any one of the appendices 1 to 5, characterized in that it is a bright spot detection device. (Note 8) The system includes a peak point exclusion unit that excludes peak points from the image information representing the input image that meet predetermined conditions from among the N peak points detected by the peak point detection unit. The mask image generation unit, By masking a predetermined range centered on the peak points that were not excluded by the peak point exclusion unit with respect to the image information representing the input image, a mask image is generated that includes M mask regions (where M is an integer between 2 and N) instead of the mask image that includes N mask regions. The antimask image generation unit, Instead of the N antimask images mentioned above, M antimask images are generated by demasking one of the M mask regions in the mask image, wherein the demasking regions differ from one another, so that when superimposed on the input image, each of the M peak points is included individually in the superimposed input image. A bright spot detection device according to any one of the appendices 1 to 7, characterized in that it is a bright spot detection device. (Note 9) The aforementioned peak point exclusion section is, Of the N peak points in the input image detected by the peak point detection unit, peak points that do not show continuity in time-series position or brightness changes in a predetermined number of frames before and after the frame relating to the input image are excluded from the image information representing the input image. The bright spot detection device described in Appendix 8, characterized by the features described above. (Note 10) The aforementioned peak point exclusion section is, By comparing multiple images with different fields of view beforehand, and recording the peak point whose position does not change even when the fields of view are different, Of the N peak points in the input image detected by the peak point detection unit, the peak points corresponding to the recorded peak points are excluded from the image information representing the input image. The bright spot detection device described in Appendix 8, characterized by the features described above. (Note 11) The aforementioned peak point exclusion section is, The input image is compared with a star map in which known stars have been pre-registered. Of the N peak points in the input image detected by the peak point detection unit, the peak points corresponding to stars registered in the star map are excluded from the image information representing the input image. The bright spot detection device described in Appendix 8, characterized by the features described above. (Note 12) It is mounted on a satellite, The bright point that the determination unit determines is a star. A bright spot detection device according to any one of the appendices 8 to 11, characterized by the above. (Note 13) A method for detecting bright spots using a bright spot detection device, The peak point detection unit performs the following steps: Based on image information representing the input image, it detects N local peak points (where N is an integer of 2 or more) contained in the input image; The mask image generation unit generates a mask image that includes N mask regions by masking a predetermined range centered on each of the detected peak points with respect to the image information. The antimask image generation unit generates N different antimask images, each of which is obtained by demasking one of the N mask regions in the mask image, and by making the mask regions to be demasked different from each other, such that when superimposed on the input image, each of the N peak points is included individually in the superimposed input image. The determination unit sequentially superimposes the N antimask images onto the input image to obtain each of the resulting superimposed input images, and determines, based on predetermined conditions, whether or not a peak point included alone in the input image is a bright spot to be detected. A method for detecting bright spots having the following characteristics. [Industrial applicability]

[0131] This disclosure makes it possible to individually determine whether each peak point is a target bright spot, even when multiple peak points are close to each other, and is suitable for use in bright spot detection devices and bright spot detection methods. [Explanation of Symbols]

[0132] 10, 10c, 10d Bright spot detection device, 11 Peak point detection unit, 12 Mask image generation unit, 13 Antimask image generation unit, 14 Judgment unit, 15 Complementary processing unit, 16 Peak point exclusion unit, 51 Processing circuit, 52 CPU, 53 Memory, 131 Distance calculation unit, 132 Antimask processing unit, P1~P3 Peak points, P11~P14 Peak points, P21~P27 Peak points, P31~P33 Peak points, P41~P42 Peak points.

Claims

1. A peak point detection unit detects N local peak points (where N is an integer of 2 or more) contained in an input image based on image information representing the input image. A mask image generation unit generates a mask image that includes N mask regions by masking a predetermined range centered on each of the detected peak points of the image information. An antimask image generation unit generates N antimask images obtained by demasking one of the N mask regions in the aforementioned mask image, wherein the mask regions to be demasked differ from one another, so that when superimposed with the input image, each of the N peak points is included individually in the superimposed input image. A determination unit determines, based on predetermined conditions, whether a peak point included alone in each of the superimposed input images obtained by sequentially superimposing the N antimask images onto the input image is a bright spot to be detected. A bright spot detection device comprising the above components.

2. The antimask image generation unit, In the aforementioned mask image, if there is a portion where multiple mask regions overlap among the N mask regions, the anti-masking processing unit includes a mechanism that removes the mask from one mask region located in the overlapping portion, specifically the portion that does not overlap with the other mask regions located in that overlapping portion. The bright spot detection device according to claim 1, characterized in that it is a bright spot detection device.

3. The mask image generation unit sets each of the N mask regions as a circular region having the same radius. The antimask image generation unit, The input image includes a distance calculation unit that calculates the distance between each of the N peak points detected, The aforementioned anti-masking processing unit is Based on the distance calculated by the distance calculation unit and the radius of the mask region, it is determined whether or not there is an overlapping portion among the N mask regions in the mask image. The bright spot detection device according to claim 2, characterized in that it is as described above.

4. The determination unit, The correlation between the luminance distribution of a peak point included alone in the input image overlaid with the aforementioned antimask image and a two-dimensional Gaussian distribution is evaluated, and based on the results of this evaluation, it is determined whether or not the peak point is a star. A bright spot detection device according to any one of claims 1 to 3.

5. The determination unit, As part of the correlation, the correlation coefficient is calculated between the luminance distribution of a peak point included alone in the input image on which the antimask image is superimposed, and a two-dimensional Gaussian distribution. If the calculated correlation coefficient is greater than or equal to a threshold, the peak point is determined to be a star; if the calculated correlation coefficient is less than the threshold, the peak point is determined to be noise. The bright spot detection device according to claim 4, characterized in that it is a bright spot detection device.

6. The determination unit, When calculating the correlation coefficient, the luminance values ​​in the mask region of the input image on which the anti-mask image is superimposed are considered invalid values ​​and are excluded from the calculation of the correlation coefficient. The bright spot detection device according to claim 5, characterized in that it is a bright spot detection device.

7. An input image superimposed with the antimask image, in which the aforementioned peak point is included alone, By rotating the input image on which the antimask image is superimposed by a predetermined angle, or by flipping the input image on which the antimask image is superimposed vertically, horizontally, or vertically and horizontally, and then compositing it with a complementary image obtained, The system includes an interpolation processing unit that interpolates the pixel values ​​of the mask region included in the input image on which the antimask image is superimposed with the pixel values ​​of the region of the interpolation image corresponding to the mask region. The determination unit, Based on the predetermined conditions, it is determined whether the peak point included alone in the image obtained by the interpolation process is a bright spot to be detected. A bright spot detection device according to any one of claims 1 to 3.

8. The system includes a peak point exclusion unit that excludes peak points from the N peak points detected by the peak point detection unit that meet predetermined conditions from the image information representing the input image. The mask image generation unit, By masking a predetermined range centered on peak points that were not excluded by the peak point exclusion unit with respect to the image information representing the input image, a mask image is generated that includes M mask regions (where M is an integer between 2 and N) instead of the mask image that includes N mask regions. The antimask image generation unit, Instead of the N antimask images mentioned above, M antimask images are generated by demasking one of the M mask regions in the mask image, wherein the demasking regions differ from one another, so that when superimposed on the input image, each of the M peak points is included individually in the superimposed input image. A bright spot detection device according to any one of claims 1 to 3.

9. The aforementioned peak point exclusion section is, Of the N peak points in the input image detected by the peak point detection unit, peak points that do not show continuity in time-series position or brightness changes in a predetermined number of frames before and after the frame relating to the input image are excluded from the image information representing the input image. The bright spot detection device according to claim 8, characterized in that it is a bright spot detection device.

10. The aforementioned peak point exclusion section is, By comparing multiple images with different fields of view beforehand, and recording the peak point whose position does not change even when the fields of view are different, Of the N peak points in the input image detected by the peak point detection unit, the peak points corresponding to the recorded peak points are excluded from the image information representing the input image. The bright spot detection device according to claim 8, characterized in that it is a bright spot detection device.

11. The aforementioned peak point exclusion section is, The input image is compared with a star map in which known stars have been pre-registered. Of the N peak points in the input image detected by the peak point detection unit, the peak points corresponding to stars registered in the star map are excluded from the image information representing the input image. The bright spot detection device according to claim 8, characterized in that it is a bright spot detection device.

12. It is mounted on a satellite, The bright point that the determination unit determines is a star. The bright spot detection device according to claim 8, characterized in that it is a bright spot detection device.

13. A method for detecting bright spots using a bright spot detection device, The peak point detection unit detects N local peak points (where N is an integer of 2 or more) contained in the input image based on image information representing the input image. The mask image generation unit generates a mask image that includes N mask regions by masking a predetermined range centered on each of the detected peak points with respect to the image information. The antimask image generation unit generates N different antimask images, each of which is obtained by demasking one of the N mask regions in the mask image, and by making the mask regions to be demasked different from each other, such that when superimposed on the input image, each of the N peak points is included individually in the superimposed input image. The determination unit sequentially superimposes the N antimask images onto the input image to obtain each of the resulting superimposed input images, and determines, based on predetermined conditions, whether or not a peak point included alone in the input image is a bright spot to be detected. A method for detecting bright spots having the following characteristics.

Images