Space target main body centroid positioning method and system

Abstract

The invention discloses a space target main body centroid positioning method and system, and belongs to the technical field of space exploration, and the method comprises the steps: extracting a target region grey-scale image Iray from an original grey-scale image, carrying out the binarization, obtaining a binary image Ibin, respectively inhibiting the star point noise in the Iray and Ibin, then rotating the target regions in the Iray and Ibin to the level of a solar wing skeleton, and carrying out the rotation of the target regions in the Iray and Ibin. A grey-scale image Igrot and a binary image Ibrot are obtained; dividing the target area into a plurality of areas along the skeleton direction of the solar wing, calculating comprehensive similarity measurement between each area and other areas according to Irot and Ibrot, and determining the area with the minimum comprehensive similarity measurement as a main body area; if the mass center of the main body area is located on the space target, taking the mass center of the main body area as the main body centroid of the space target; and otherwise, taking the centroid of the space target in the Ibin as the main body centroid of the space target. According to the invention, accurate and stable main body centroid positioning can be carried out on space targets with various shapes and various postures in a complex imaging environment.

Description

Technical Field

[0001] The present invention belongs to the technical field of space exploration, and more specifically, relates to a method and system for centroid positioning of a spatial target body. Background Art

[0002] Centroid positioning of a spatial target body is an important part of key technologies such as satellite imaging, target recognition, and space tracking. It is mainly used to determine the central position of the spatial target body, providing precise support for operations such as attitude recognition, trajectory analysis, and rendezvous and docking. In practical applications, due to factors such as the diverse shapes and postures of spatial targets, variable lighting conditions in the environment, and different imaging distances, accurately and stably extracting the centroid of the target body faces many challenges.

[0003] Traditional centroid positioning methods usually rely on the extraction of special endpoints, lines, and geometric figures (such as circles, ellipses) in image processing. For example, "A Method for Centroid Positioning of a Spatial Target" (Patent Application No. CN202210763578.7, Publication No. CN114998437A) discloses a method for centroid positioning of a spatial target. The implementation of the method includes: obtaining the circumscribed rectangle of the target and separating it from the background, then determining the sailboard area through edge detection and fitting, separating the target body using the demarcation line, and finally determining the minimum circumscribed rectangle and centroid of the body based on the principal axis method. Such methods have good centroid positioning effects on double-wing satellites under good imaging conditions, but there are also the following three limitations: (1) Lack of generalization ability for satellites with different shapes. Traditional methods are usually only applicable to double-wing symmetric satellites, and the centroid positioning accuracy for single-wing satellites decreases significantly; (2) Insufficient consideration of noise and blur effects caused by lighting. Satellite images are often interfered by complex lighting and optical system noise. These problems not only blur the target boundary but may also generate false feature points (such as star points), seriously affecting the accuracy of centroid positioning; (3) High computational complexity and difficulty in meeting real-time requirements. The centroid positioning method based on deep learning realizes the centroid positioning of the spatial target body with the help of a deep learning model. Although its adaptability is enhanced compared with traditional centroid positioning methods, due to the large model size and high dependence on hardware resources, it is difficult to meet the requirements of actual space tasks in terms of real-time performance and resource consumption. These problems severely limit the application of existing spatial target centroid positioning methods in complex actual scenarios.

[0004] Therefore, developing a method that can accurately and stably perform centroid positioning of the main body of spatial targets with diverse shapes and variable postures in a complex imaging environment is still a technical problem and a key research direction for achieving high-precision spatial target analysis. Summary of the Invention

[0005] In view of the deficiencies of the prior art and the improvement requirements, the present invention provides a method and system for centroid positioning of a spatial target body, aiming to achieve accurate and stable centroid positioning of spatial targets with diverse shapes and variable postures in a complex imaging environment.

[0006] To achieve the above object, according to one aspect of the present invention, there is provided a method for centroid positioning of a spatial target body, including: S1: Extract a target area containing the spatial target from the originally acquired grayscale image I ori to obtain a grayscale image of the target area I gray ; S2: Binarize the grayscale image of the target area I gray to obtain a binary image I bin , and suppress the star point noise in the grayscale image of the target area I gray and the binary image I bin respectively; S3: Rotate the target areas in the grayscale image of the target area after suppressing the star point noise I gray and the binary image I bin to the horizontal of the solar wing skeleton to obtain a grayscale image with normalized image plane posture I g_rot and a binary image I b_rot ; S4: Divide the target area along the direction of the solar wing skeleton into N regions. For each region z, calculate the sum of the differences in grayscale values between it and the remaining regions according to the grayscale image I g_rot , and calculate the sum of the binary differences between it and the remaining regions according to the binary image D g , and calculate the comprehensive similarity measure between region z and the remaining regions according to I b_rot D b . Determine the region with the minimum comprehensive similarity measure D D as the main body region of the spatial target; N > 1, and are weights and they are not both 0 at the same time; ​​S5: Determine whether the centroid of the main region is located on the space target. If so, use the centroid of the main region as the main centroid of the space target P ; Otherwise, use the centroid of the space target in the binary image after suppressing star noise I bin as the main centroid of the space target P 。

[0007] Furthermore, in S2, the method for suppressing star noise includes: Extract each contour in the target region and calculate the area of the region enclosed by each contour; Determine the regions with an area smaller than a preset area threshold in the regions enclosed by each contour as star points and filter them out.

[0008] Furthermore, in S2, the method for suppressing stars also includes: After determining the regions with an area smaller than a preset area threshold in the regions enclosed by each contour as star points and filtering them out, calculate the minimum circumscribed circle of the remaining regions, and calculate the ratio of the region area to the area of the minimum circumscribed circle as the filling rate of the corresponding region; Determine the regions with a filling rate greater than a preset filling rate threshold and a minimum circumscribed circle radius smaller than a preset roundness threshold in the remaining regions as star points and filter them out.

[0009] Furthermore, in S3, rotate the grayscale image I gray of the target region after suppressing star noise I bin and the binary image so that the target regions in both are rotated to the horizontal of the solar wing skeleton, including: ； Extract the solar wing skeleton of the space target and calculate the angle between the extracted solar wing skeleton and the horizontal direction According to the angle I gray rotate the target regions in the grayscale image I bin of the target region after suppressing star noise and the binary image

[0010] respectively, until the solar wing skeleton in the target region is horizontal. Furthermore, extracting the solar wing skeleton of the space target includes: I bin Perform line detection on the binary image after suppressing star noise L 1 ; Remove the line L 1After other collinear straight lines, find the line segment with the longest length among the remaining line segments L 2 ; In the binary image after suppressing the star point noise I bin Among them, using the average slope of the line segment L 1 And the line segment L 2 As the slope, draw a straight line L 3 Passing through the centroid of the target area, as the solar wing skeleton

[0011] Further, S1 includes: S11: Perform Gaussian blur processing and adaptive histogram equalization with limited contrast on the originally collected grayscale image I ori In sequence to obtain the preprocessed image I pre ; S12: Perform threshold adaptive binarization operation on the preprocessed image I pre ; S13: After performing opening and closing operations on the preprocessed image after binarization operation I pre Perform dilation operation to obtain the image I pre ' ; S14: Search for connected regions in the image And select the connected region with the largest area as the target area, extract the target area from the preprocessed image I pre To obtain the target area grayscale image I gray .

[0012] Further, in S4, the gray value difference is measured by the mean absolute error, and the binary difference is measured by the overlap coefficient

[0013] According to another aspect of the present invention, there is provided a computer program product, including a computer program; when the computer program is executed by a processor, it implements the above-mentioned spatial target main body centroid positioning method provided by the present invention

[0014] According to another aspect of the present invention, there is provided a computer-readable storage medium, including a stored computer program; when the computer program is executed by a processor, it controls the device where the computer-readable storage medium is located to execute the above-mentioned spatial target main body centroid positioning method provided by the present invention

[0015] According to another aspect of the present invention, there is provided a centroid positioning system for a spatial target main body, comprising: A computer-readable storage medium for storing a computer program; And a processor for reading the computer program stored in the computer-readable storage medium to implement the above-mentioned centroid positioning method for the spatial target main body provided by the present invention.

[0016] Generally speaking, through the above technical solutions conceived by the present invention, the following beneficial effects can be achieved: (1) After extracting the grayscale image of the target area from the original grayscale image, the present invention will perform star point suppression on it, thereby being able to suppress the interference of environmental noise on the centroid positioning result of the spatial target main body; on this basis, the present invention further performs image plane attitude normalization processing on the target area image, thereby being able to suppress the interference of different attitudes of the spatial target on the positioning result. Then, the target area is divided along the direction of the solar wing skeleton, and the comprehensive similarity measure between each area and other areas is calculated based on the grayscale image and the binary image, thereby determining the main body area. Based on the characteristic that the difference between the main body area of the spatial target and the remaining areas is the largest, the present invention can accurately locate the main body area for spatial targets of various shapes and various attitudes, and then accurately locate the centroid of the main body area.

[0017] (2) In the preferred solution of the present invention, the method for suppressing star point noise in the target area includes extracting each contour in the target area and calculating the area of the region enclosed by each contour, and determining the regions with smaller areas as star points for filtering; in the actual space detection scenario, since stars are usually far away from the spatial target, the corresponding areas in the target area are small. Therefore, the present invention can accurately filter out the star points in the target area and avoid the influence of the bright star points in the target area on the centroid positioning result of the spatial target main body. In its further preferred solution, the present invention will calculate the minimum circumscribed circle for the remaining regions respectively, and calculate the filling rate and the radius of the minimum circumscribed circle, and determine the regions with larger filling rates and smaller minimum circumscribed circle areas as star points for filtering; due to the reason of light emission, the detected light spots of stars are usually circular or approximately circular, making the filling rate (the ratio of the area of the region enclosed by the contour to the minimum circumscribed circle) of the corresponding region larger, and due to the long distance, the radius of its minimum circumscribed circle is smaller. Therefore, the present invention determines and filters out the remaining star points in the target area according to the filling rate and the radius of the minimum circumscribed circle, can more accurately filter out the remaining star points in the target area, further suppress the influence of star points on the centroid positioning result of the spatial target main body, and improve the positioning accuracy.

[0018] (3) By extracting the solar panel skeleton of the space target, the present invention determines the rotation angle for achieving attitude normalization of the image plane, and the related process is not affected by the shape and attitude of the space target. In a further preferred embodiment thereof, the extraction of the solar panel skeleton of the space target is completed by means of line detection. The related extraction process is consistent with the structural features of the space target and is not affected by the shape and attitude of the space target, having a high extraction accuracy.

[0019] (4) In a further preferred embodiment of the present invention, before extracting the gray-scale image of the target area from the originally acquired gray-scale image, the star point noise in the original gray-scale image is preliminarily suppressed by a denoising algorithm, thereby further suppressing the influence of the star point noise in the environment on the positioning result and improving the accuracy of the centroid positioning of the space target main body. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 It is a flowchart of the method for centroid positioning of the space target main body provided by an embodiment of the present invention.

[0021] Figure 2 It is the originally acquired gray-scale image provided by an embodiment of the present invention.

[0022] Figure 3 It is a schematic diagram of the centroid positioning result of the space target main body provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0023] In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0024] In the present invention, terms such as "first" and "second" in the present invention and the drawings (if any) are used to distinguish similar objects and do not necessarily need to describe a specific order or sequence.

[0025] In spatial imaging, different types of targets usually exhibit significant shape differences. The same target may also show large morphological changes due to perspective effects under different postures. In addition, with the change of imaging distance and the interference of imaging noise, the details of the target may be blurred, and the boundaries of the target contour may be difficult to accurately identify. At the same time, complex spatial illumination conditions may also cause problems such as uneven brightness or blurred edges of the target in the image. The combined effects of these factors make it difficult for traditional centroid localization methods based on geometric features such as points and lines to operate stably under variable conditions, while centroid localization methods based on deep learning rely on large-scale deep learning models and are difficult to meet the requirements of actual spatial tasks in terms of real-time performance and resource consumption. To solve the above problems, the present invention provides a method and system for locating the centroid of the main body of a spatial target. The overall concept is to locate the main body area by using the similarity and difference degrees between regions of the spatial target, and then locate the centroid of the main body of the spatial target based on the location result of the main body area, avoiding dependence on the shape and posture of the spatial target, and fully considering the influence of star point noise on the centroid localization result of the spatial target. Before locating the main body area, the relevant star point noise is effectively filtered, so as to achieve accurate and stable centroid localization of spatial targets with diverse shapes and variable postures in complex environments.

[0026] The following are embodiments.

[0027] A method for locating the centroid of the main body of a spatial target, as Figure 1 shown, includes: S1: Extract the target area containing the spatial target from the originally acquired grayscale image I ori to obtain a grayscale image of the target area I gray ; S2: Binarize the grayscale image of the target area I gray to obtain a binary image I bin , and suppress the star point noise in the grayscale image of the target area I gray and the binary image I bin respectively; S3: Rotate the target areas in the grayscale image of the target area after suppressing star point noise I gray and the binary image I bin to the horizontal of the solar wing skeleton, and obtain a grayscale image I g_rot with normalized image plane posture and a binary image I b_rot ; S4: Divide the target area along the direction of the solar wing skeleton into N regions, for each region z, according to the grayscale image I g_rot Calculate the sum of the gray value differences between it and the rest of the regions D g , and according to the binary graph I b_rot Calculate the sum of the binary differences between it and the rest of the regions D b , and follow Calculate the comprehensive similarity measure between region z and the rest of the regions D , the comprehensive similarity measure D The smallest area is determined as the main area of the space target; N >1, and is the weight and is 0 when the two are different; S5: Determine whether the centroid of the main area is located on the space target. If so, take the centroid of the main area as the main centroid of the space target. P Otherwise, the binary image after suppressing the star point noise I bin The center of mass of the space target is used as the main centroid of the space target P .

[0028] The following is a detailed explanation of the specific implementation of each step.

[0029] Figure 2 The figure shows an example of an originally collected grayscale image, which contains a large amount of background information. On this basis, by determining the target area, the subsequent steps can focus on the target area and avoid redundant calculations of the entire image, thereby speeding up the execution of subsequent processing steps, meeting real-time requirements, and eliminating the interference of irrelevant information, thereby improving the positioning accuracy of the centroid of the main body of the space target. Therefore, in step S1 of this embodiment, the target area containing the space target will first be extracted from the originally collected grayscale image.

[0030] During the imaging process, due to the interference of random noises such as Gaussian noise and white noise, there may be high-brightness stray light spots or blurred areas in the image, making the target edges unclear and the details difficult to identify. In addition, there are a large number of high-brightness star points in the space environment. These noises and blurs not only affect the image quality but also pose great challenges to subsequent centroid localization. Therefore, it is necessary to perform preprocessing to improve the image quality and the robustness of subsequent algorithms. Based on these considerations, in step S1 of this embodiment, before extracting the grayscale image of the target area from the originally acquired grayscale image, preprocessing will be carried out first. During the preprocessing process, the high-brightness star points and noises are preliminarily suppressed through a denoising algorithm, and at the same time, the technology of sharpening to enhance the edges is used to make the target contour clearer, thus providing a better basis for accurately and stably locating the centroid of the main body.

[0031] Specifically, S1 of this embodiment includes: S11: Perform Gaussian blur processing on the originally acquired grayscale image I ori in sequence to reduce noises and star points; perform adaptive histogram equalization with limited contrast on the Gaussian-blurred image to enhance the contrast and detail effect of the target area and obtain a preprocessed image I pre ; S12: Perform a binarization operation with adaptive threshold on the preprocessed image I pre ; S13: Perform opening and closing operations on the preprocessed image after the binarization operation I pre to further suppress the high-brightness star point noises and fill the holes in the binary area, making the target area smoother and more complete, and then perform a dilation operation to connect adjacent target areas, thereby obtaining an image I pre ' ; S14: Search for connected regions in the image and select the connected region with the largest area as the target area, and extract the target area from the preprocessed image I pre to obtain the grayscale image of the target area I gray 。

[0032] In spatial imaging, the complexity of the environment has a significant impact on the detection and positioning of targets. In addition, there are a large number of bright star points in the spatial environment, which are likely to interfere with subsequent target positioning steps. Since the brightness values of these star points are similar to those of the target area, they may be misidentified as the features of the target, resulting in deviations in centroid positioning. In step S1 of this embodiment, although the star point noise is suppressed to a certain extent through a denoising algorithm, the star points with brightness values similar to those of the target area cannot be suppressed. Therefore, in step S2 of this embodiment, based on the shape features of the contours of these star point noises and the areas of the enclosed regions, a star point noise suppression method that can effectively suppress the star points in the target area is further proposed, which specifically includes: Extract each contour in the target area and calculate the area of the region enclosed by each contour; among them, the extraction of the contour in the target area can be realized on the basis of performing an adaptive threshold binary operation on the grayscale image of the target area I gray operation. Generally, the stars are far from the spatial target. Therefore, in the target area, the area of the region is small. Based on this, in this embodiment, the regions with areas smaller than the preset area threshold in the regions enclosed by each contour are determined as star points and filtered out.

[0033] This embodiment further considers that due to the reason of luminescence, the light spots formed by the stars in the image are circular or approximately circular. Therefore, the method for suppressing star points proposed in this embodiment also includes: after determining the regions with areas smaller than the preset area threshold in the regions enclosed by each contour as star points and filtering them out, calculate the minimum circumscribed circle of the remaining regions, and calculate the ratio of the region area to the area of the minimum circumscribed circle as the filling rate of the corresponding region; Determine the regions with filling rates greater than the preset filling rate threshold and minimum circumscribed circle radii less than the preset roundness threshold in the remaining regions as star points and filter them out.

[0034] It is easy to understand that when suppressing the star points in the target area, the relevant area threshold, filling rate threshold, and roundness threshold can be determined according to the environmental characteristics in the current space detection scenario.

[0035] In the processing of spatial targets with complex postures, the shape of the target may change significantly due to the perspective effect. To improve the accuracy of recognition, it is usually necessary to perform normalization processing on the target, that is, to perform standardized adjustment of the posture of the target image, so as to facilitate the recognition and comparison of different regions. Based on this, in this embodiment, through step S3, before determining the main area of the spatial target, the image will be processed by image plane normalization. Specifically, in S3, the grayscale image of the target area after suppressing star point noise I gray and the binary image I binThe target regions in Extract the solar wing framework of the space target and calculate the angle between the extracted solar wing framework and the horizontal direction ； According to the angle For the grayscale image of the target region after suppressing star point noise I gray and the binary image I bin rotate the target regions therein until the solar wing framework in the target regions is horizontal.

[0036] Extract the solar wing framework of the space target, specifically including: Perform line detection on the binary image after suppressing star point noise I bin and select the longest line segment from the detected line segments L 1 ; After removing other lines collinear with the line segment L 1 find the longest line segment among the remaining line segments L 2 ; In the binary image after suppressing star point noise I bin draw a line L 1 with the average slope of the line segment L 2 and passing through the centroid of the target region L 3 as the solar wing framework.

[0037] The different regions of the target show significant differences in the grayscale image and the binary image. Step S4 of this embodiment utilizes these differences to effectively locate the main region of the space target. To accurately measure the differences between different regions in the space target, this embodiment comprehensively calculates the grayscale difference and binary difference between regions through the grayscale image and the binary image. Optionally, in this embodiment, the normalized mean absolute error (MAE) is used as the measure of the grayscale value difference. A lower MAE value indicates that the two regions are more similar. The calculation formula is as follows: ; where G 1 and G 2 represent two regions in the grayscale image, and are the grayscale values of the two regions at the th pixel point,M is the total number of pixels in the region.

[0038] Optionally, in this embodiment, the normalized Overlap Coefficient (OC) is used as a measure of binary difference. OC measures the similarity by counting the proportion of pixels with a value of 255 (i.e., white) at the same position in two regions. The higher the OC value, the more similar the two regions are. The calculation formula is as follows:

[0039] where, B 1 and B 2 represent two regions in the binary image; and are the binary pixel values (taking 0 or 255) of the two regions at the th pixel point, M is the total number of pixels in the region.

[0040] For each region z, calculate the mean absolute error and overlap coefficient between this region and each of the remaining regions respectively. Normalize the obtained mean absolute error and overlap coefficient so as to compare them on the same scale. Then add the normalized mean absolute errors to obtain D g , add the normalized overlap coefficients to obtain D b ; Optionally, in this embodiment, when calculating the comprehensive similarity measure of the region, the weight of the normalized mean absolute error is set to -0.5, and the weight of the normalized overlap coefficient is set to 0.5, that is D =-0.5 D g +0.5 D b , and this comprehensive similarity measure is used to evaluate the overall similarity degree between region z and the remaining regions. It should be noted that when calculating the comprehensive similarity measure of the region, the setting of the weight can be adjusted according to the actual space detection scenario. In some specific cases, one of the weights can be set to 0, but the two cannot be set to 0 at the same time.

[0041] After calculating the comprehensive similarity degree of each region in this embodiment, based on the characteristic that the difference between the main region and the remaining regions is the largest, the region with the smallest comprehensive similarity degree is determined as the main region of the space target, which can accurately locate the main region in the space target.

[0042] For some space targets with specific shapes, the centroid of their main regions may not be located on the space target. In response to this situation, this embodiment will further judge and correct it to ensure that the located main centroid is on the target.

[0043] For Figure 2 the grayscale image shown, the centroid positioning result of the spatial target main body finally obtained in this embodiment is as Figure 3 shown.

[0044] In this embodiment, the typical two-wing target Deepspace is further used for error testing. The imaging distance is set to 10 kilometers, and the average error in each attitude is within 2 pixels. The result of this error analysis shows that in this embodiment, accurate and stable centroid positioning of spatial targets with diverse shapes and variable postures can also be achieved in complex environments.

[0045] Embodiment 2: A computer program product includes a computer program; when the computer program is executed by a processor, it implements the spatial target main body centroid positioning method provided in the above Embodiment 1.

[0046] Embodiment 3: A computer-readable storage medium includes a stored computer program; when the computer program is executed by a processor, it controls the device where the computer-readable storage medium is located to execute the spatial target main body centroid positioning method provided in the above Embodiment 1.

[0047] Embodiment 4: A spatial target main body centroid positioning system includes: A computer-readable storage medium for storing a computer program; and a processor for reading the computer program stored in the computer-readable storage medium to implement the spatial target main body centroid positioning method provided in the above Embodiment 1.

[0048] It is easy for those skilled in the art to understand that the above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, and improvements made within the spirit and principles of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for centroid positioning of a spatial target body, characterized in that, Including: S1: Extract the target region containing the spatial target from the originally collected grayscale image to obtain the grayscale image of the target region I ori ; I gray ; S2: Binarize the grayscale image of the target region I gray to obtain a binary image I bin , and suppress the star noise in the grayscale image of the target region I gray and the binary image I bin respectively; S3: Rotate the grayscale image of the target area after suppressing the star point noise I gray and the binary image I bin of the target area to the horizontal of the solar wing skeleton, and obtain the grayscale image I g_rot and the binary image I b_rot ; S4: Divide the target area along the direction of the solar panel framework into N zones. For each zone z, calculate the sum of the grayscale value differences between it and the other zones according to the grayscale image I g_rot , and calculate the sum of the binary differences between it and the other zones according to the binary image D g . Then, calculate the comprehensive similarity measure between zone z and the other zones according to I b_rot . Determine the zone with the smallest comprehensive similarity measure as the main body area of the spatial target; D b , where > 1, D and D are weights and they are not both 0 at the same time; N > 1, and are weights and they are not both 0 at the same time; S5: Determine whether the centroid of the main area is located on the spatial target. If so, use the centroid of the main area as the main centroid of the spatial target P ; Otherwise, the binary image after suppressing the star point noise I bin The centroid of the spatial target in it is used as the main centroid of the spatial target P .

2. The centroid positioning method for the spatial target main body according to claim 1, characterized in that In S2, the methods for suppressing star noise include: Extract each contour in the target area and calculate the area of the region enclosed by each contour; In the regions enclosed by each contour, the regions with an area smaller than the preset area threshold are determined as star points and filtered out.

3. The centroid positioning method for the spatial target main body according to claim 2, characterized in that, In S2, the method for suppressing star points also includes: after determining that the regions with an area smaller than the preset area threshold in the regions enclosed by each contour are star points and filtering them out, calculate the minimum circumscribed circle of the remaining regions, and calculate the ratio of the region area to the area of the minimum circumscribed circle as the filling rate of the corresponding region; In the remaining regions, the regions with a filling rate greater than the preset filling rate threshold and a minimum circumscribed circle radius smaller than the preset roundness threshold are determined as star points and filtered out.

4. The centroid positioning method of the spatial target main body according to any one of claims 1 to 3, characterized in that In S3, the grayscale image of the target area after suppressing the star point noise I gray and the binary image I bin in which the target areas are both rotated to the level of the solar wing skeleton, including: Extract the solar panel skeleton of the space target and calculate the angle between the extracted solar panel skeleton and the horizontal direction ； According to the included angle Rotate the grayscale image of the target area after suppressing star point noise I gray and the binary image I bin in the target area until the solar wing skeleton in the target area is horizontal.

5. The centroid positioning method of the spatial target main body according to claim 4, characterized in that, Extracting the solar wing skeleton of the space target includes: Binary image after suppressing star point noise I bin Perform line detection on it, and select the line segment with the longest length from the detected line segments L 1 ; Remove other lines collinear with the line segment L 1 After that, find the line segment with the longest length among the remaining line segments L 2 ; Binary image after suppressing star noise I bin Among them, using the average slope of the line segment L 1 and the line segment L 2 as the slope, draw a straight line L 3 passing through the centroid of the target area as the solar wing skeleton.

6. The centroid positioning method of the spatial target main body according to any one of claims 1 to 3, characterized in that S1 includes: S11: Perform Gaussian blur processing and adaptive histogram equalization with limited contrast on the originally collected grayscale image I ori in sequence to obtain a preprocessed image I pre ; S12: Perform an adaptive threshold binary operation on the preprocessed image I pre ; S13: The preprocessed image after binarization operation I pre After performing opening and closing operations, perform a dilation operation to obtain an image I pre ' ; S14: Search for connected components in the image and select the connected component with the largest area as the target region. Extract the target region from the preprocessed image I pre to obtain the grayscale image of the target region I gray .

7. The centroid positioning method of the spatial target main body according to any one of claims 1 to 3, characterized in that In S4, the gray value difference is measured by the normalized mean absolute error, and the binary difference is measured by the normalized overlap coefficient.

8. A computer program product, characterized in that, Including a computer program; when the computer program is executed by a processor, it implements the space target centroid positioning method according to any one of claims 1 to 7.

9. A computer-readable storage medium, characterized in that, Including a stored computer program; when the computer program is executed by a processor, it controls the device where the computer-readable storage medium is located to execute the space target centroid positioning method according to any one of claims 1 to 7.

10. A centroid positioning system for a spatial target main body, characterized in that Including: A computer-readable storage medium for storing a computer program; And a processor for reading the computer program stored in the computer-readable storage medium and implementing the space target centroid positioning method according to any one of claims 1 to 7.

Images