A method and system for locating the centroid of a space target

Abstract

The present invention discloses a method and system for locating the centroid of a space target, which belongs to the field of space detection technology, including: extracting a target area grayscale image from an original grayscale image; I gray And perform binarization to obtain a binary image I bin , respectively inhibit I gray and I bin The star point noise in I gray and I bin The target areas in the image are rotated to the level of the solar wing skeleton to obtain the grayscale image. I g_rot and binary images I b_rot ; Divide the target area into multiple areas along the direction of the solar wing frame, according to I g_rot and I b_rot Calculate the comprehensive similarity measure between each region and the rest of the regions, and determine the region with the smallest comprehensive similarity measure as the main region; if the centroid of the main region is located on the spatial target, the centroid of the main region is used as the main centroid of the spatial target; otherwise, I bin The centroid of the space target is used as the main centroid of the space target. The present invention can accurately and stably locate the main centroid of space targets with various shapes and postures in complex imaging environments.

Description

Technical Field

[0001] The present invention belongs to the field of space detection technology, and more specifically, relates to a method and system for locating the centroid of a space target body. Background Art

[0002] Centroid localization of space targets is a crucial step in key technologies such as satellite imaging, target recognition, and space tracking. It is primarily used to determine the center position of a space target, providing precise support for gesture recognition, trajectory analysis, and rendezvous and docking operations. In practical applications, accurately and stably extracting the centroid of a target poses numerous challenges due to factors such as the diverse shapes and attitudes of space targets, variable lighting conditions in the environment, and varying imaging distances.

[0003] Traditional centroid positioning methods usually rely on the extraction of special endpoints, lines, and geometric shapes (such as circles and ellipses) in image processing. For example, "A method for centroid positioning of space targets" (patent application number CN202210763578.7, publication number CN114998437A) discloses a method for centroid positioning of space targets, wherein the method implementation includes: obtaining the bounding 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 dividing line, and finally determining the minimum bounding rectangle and centroid of the body based on the principal axis method. This type of method has a good effect on centroid positioning of dual-wing satellites under good imaging conditions, but it also has the following three limitations: (1) Lack of generalization processing capabilities for satellites of different shapes. Traditional methods are usually only applicable to dual-wing symmetrical satellites, and the centroid positioning accuracy of single-wing satellites is significantly reduced; (2) Insufficient consideration of noise and blurring effects caused by illumination. Satellite images are often affected by complex lighting and optical system noise. These problems not only blur the target boundary, but also may generate false feature points (such as stars), which seriously affect the accuracy of centroid positioning. (3) The computational complexity is high and it is difficult to meet real-time requirements. The centroid positioning method based on deep learning uses a deep learning model to realize the centroid positioning of the main body of the space target. Although its adaptability is enhanced compared with the traditional centroid positioning method, due to the large model size and high dependence on hardware resources, it is difficult to meet the requirements of actual space missions in terms of real-time performance and resource consumption. These problems seriously limit the application of existing space target centroid positioning methods in complex actual scenarios.

[0004] Therefore, developing a method that can accurately and stably locate the centroid of space targets with diverse shapes and changing postures in complex imaging environments remains a technical challenge and key research direction for achieving high-precision space target analysis. Summary of the Invention

[0005] In response to the defects of the existing technology and the need for improvement, the present invention provides a method and system for locating the centroid of a space target body, the purpose of which is to achieve accurate and stable centroid positioning of space targets with diverse shapes and changing postures in a complex imaging environment.

[0006] To achieve the above object, according to one aspect of the present invention, a method for locating the centroid of a space target body is provided, comprising:

[0007] S1: Grayscale image collected from the original I ori Extract the target area containing the spatial target and obtain the grayscale image of the target area I gray ;

[0008] S2: Grayscale image of target area I gray Binarize to get a binary image I bin , and suppress the target area grayscale image respectively I gray and binary images I bin Star point noise in

[0009] S3: Grayscale image of the target area after suppressing star point noise I gray and binary images I bin The target areas in the image are rotated to the level of the solar wing skeleton to obtain the grayscale image with normalized image plane posture. I g_rot and binary images I b_rot ;

[0010] 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;

[0011] S5: Determine whether the centroid of the main area is located on the space target. If so, the centroid of the main area is used as the main centroid of the space target. P Otherwise, the binary image after suppressing the star point noise will be I bin The center of mass of the space target is used as the main centroid of the space target P .

[0012] Furthermore, in S2, the star point noise suppression method includes:

[0013] Extract each contour in the target area and calculate the area of the region enclosed by each contour;

[0014] In the areas enclosed by each contour, the areas whose area is smaller than the preset area threshold are determined as star points and filtered out.

[0015] Furthermore, in S2, the method of suppressing star points further includes: determining, among the areas enclosed by the contours, areas whose areas are smaller than a preset area threshold as star points, filtering them out, calculating the minimum circumscribed circle of each remaining area, and calculating the ratio of the area of the area to the area of the minimum circumscribed circle as the fill rate of the corresponding area;

[0016] In the remaining area, the areas whose filling rate is greater than a preset filling rate threshold and whose minimum circumscribed circle radius is less than a preset roundness threshold are determined as star points and filtered out.

[0017] Furthermore, in S3, the grayscale image of the target area after suppressing the star point noise is I gray and binary images I bin The target areas in the image are all rotated to the level of the solar wing frame, including:

[0018] Extract the solar wing frame of the space target and calculate the angle between the extracted solar wing frame and the horizontal direction ；

[0019] According to the angle The grayscale image of the target area after suppressing star noise is I gray and binary images I bin The target area in the image is rotated until the solar wing frame in the target area is level.

[0020] Furthermore, the solar wing skeleton of the space target is extracted, including:

[0021] Binary image after suppressing star noise I binPerform straight line detection on the line segment and select the longest line segment from the detected line segments L 1 ;

[0022] Remove and line segments L 1 After finding the other collinear lines, find the longest line segment among the remaining line segments L 2 ;

[0023] Binary image after suppressing star noise I bin In the line segment L 1 and line segments L 2 The average slope of the slope is the slope, and a straight line is drawn through the centroid of the target area. L 3 , as the solar wing skeleton.

[0024] Furthermore, S1 includes:

[0025] S11: Grayscale image of the original collection I ori Gaussian blur processing and adaptive histogram equalization with limited contrast are performed in sequence to obtain the preprocessed image I pre ;

[0026] S12: Preprocessing images I pre Perform threshold adaptive binarization operation;

[0027] S13: Preprocessing image after binarization operation I pre After performing the opening and closing operations, the expansion operation is performed to obtain the image I pre ' ;

[0028] S14: In the image Search for connected domains in the image and select the connected domain with the largest area as the target area. I pre Extract the target area and get the grayscale image of the target area I gray .

[0029] Furthermore, in S4, the gray value difference is measured using the mean absolute error, and the binary value difference is measured using the overlap coefficient.

[0030] According to another aspect of the present invention, a computer program product is provided, comprising a computer program; when the computer program is executed by a processor, the above-mentioned method for locating the centroid of a space target body provided by the present invention is implemented.

[0031] According to another aspect of the present invention, a computer-readable storage medium is provided, comprising a stored computer program; when the computer program is executed by a processor, the device where the computer-readable storage medium is located is controlled to execute the above-mentioned method for locating the centroid of a space target body provided by the present invention.

[0032] According to another aspect of the present invention, there is provided a space target body centroid positioning system, comprising:

[0033] a computer-readable storage medium for storing a computer program;

[0034] and a processor for reading a computer program stored in a computer-readable storage medium to implement the above-mentioned method for locating the centroid of a space target body provided by the present invention.

[0035] In general, the above technical solutions conceived by the present invention can achieve the following beneficial effects:

[0036] (1) After extracting the grayscale image of the target area from the original grayscale image, the present invention will suppress the star points on it, thereby suppressing the interference of environmental noise on the positioning result of the centroid of the main body of the space target; on this basis, the target area image is further normalized in the image plane, thereby suppressing the interference of different postures of the space target on the positioning result, and then the target area is divided into regions along the direction of the solar wing skeleton, and the comprehensive similarity measurement between each region and other regions is calculated based on the grayscale image and the binary image, thereby determining the main region. Based on the characteristic that the main region of the space target has the greatest difference from the remaining regions, the present invention can accurately locate the main region of space targets of various shapes and postures, and then accurately locate the centroid of the main region.

[0037] (2) In the preferred embodiment of the present invention, the method of suppressing the star point noise in the target area includes extracting each contour in the target area, calculating the area of the area enclosed by each contour, and determining the area with a smaller area as a star point for filtering out. In the actual space detection scenario, since stars are usually far away from the space target, the corresponding area in the target area is small. Therefore, the present invention can accurately filter out the star points in the target area and avoid the influence of the highlighted star points in the target area on the centroid positioning result of the space target body. In its further preferred embodiment, the present invention calculates the minimum circumscribed circle for the remaining areas respectively, and calculates the filling rate and the minimum circumscribed circle radius, and determines the areas with a larger filling rate and a smaller minimum circumscribed circle area as star points for filtering out; due to the light emission of stars, their detected light spots are usually circular or approximately circular, so that the filling rate of the corresponding area (the ratio of the area enclosed by the outline to the minimum circumscribed circle) is larger, and due to the farther distance, the minimum circumscribed circle radius is smaller. Therefore, the present invention determines and filters out the star points in the remaining areas according to the filling rate and the minimum circumscribed circle radius, which can more accurately filter out the remaining star points in the target area, further suppress the influence of the star points on the centroid positioning results of the space target body, and improve the positioning accuracy.

[0038] (3) The present invention determines the rotation angle for achieving normalized image plane attitude by extracting the solar wing skeleton of the space target. The related process is not affected by the shape and attitude of the space target. In a further preferred embodiment, the solar wing skeleton of the space target is extracted by line detection. The related extraction process is consistent with the structural characteristics of the space target and is not affected by the shape and attitude of the space target, thereby achieving high extraction accuracy.

[0039] (4) In a further preferred embodiment of the present invention, before extracting the target area grayscale image from the originally collected grayscale image, the star point noise in the original grayscale 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 body. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 This is a flow chart of a method for locating the centroid of a space target provided by an embodiment of the present invention.

[0041] Figure 2 This is the originally collected grayscale image provided by the embodiment of the present invention.

[0042] Figure 3 A schematic diagram of the centroid positioning results of a space target provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0043] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely for the purpose of explaining the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.

[0044] In the present invention, the terms "first", "second", etc. (if any) in the present invention and the drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence.

[0045] In spatial imaging, different types of objects often exhibit significant shape differences. The same object can also exhibit significant morphological variations in different poses due to perspective effects. Furthermore, as imaging distance changes and imaging noise interferes, object details can become blurred, and the boundaries of the object's outline can be difficult to accurately identify. Furthermore, complex spatial lighting conditions can also cause uneven brightness or blurred edges on objects in images. The combined effects of these factors make traditional centroid localization methods based on geometric features such as points and lines difficult to operate stably under changing conditions. Deep learning-based centroid localization methods, which rely on large-scale deep learning models, struggle to meet the real-time and resource consumption requirements of practical spatial tasks. To solve the above problems, the present invention provides a method and system for locating the main centroid of a space target. The overall concept is to locate the main area by utilizing the similarities and differences between the various areas of the space target, and then locate the main centroid of the space target based on the main area positioning results, avoiding dependence on the shape and posture of the space target, and fully considering the influence of star point noise on the centroid positioning results of the space target. Before locating the main area, the relevant star point noise is effectively filtered out, so that accurate and stable main centroid positioning of space targets with diverse shapes and changeable postures can be achieved even in complex environments.

[0046] The following are examples.

[0047] A method for locating the centroid of a space target, such as Figure 1 Shown, including:

[0048] S1: Grayscale image collected from the original I ori Extract the target area containing the spatial target and obtain the grayscale image of the target area I gray ;

[0049] S2: Grayscale image of target area I gray Binarize to get a binary image I bin, and suppress the target area grayscale image respectively I gray and binary images I bin Star point noise in

[0050] S3: Grayscale image of the target area after suppressing star point noise I gray and binary images I bin The target areas in the image are rotated to the level of the solar wing skeleton to obtain the grayscale image with normalized image plane posture. I g_rot and binary images I b_rot ;

[0051] 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;

[0052] S5: Determine whether the centroid of the main area is located on the space target. If so, the centroid of the main area is used as the main centroid of the space target. P Otherwise, the binary image after suppressing the star point noise will be I bin The center of mass of the space target is used as the main centroid of the space target P .

[0053] The following is a detailed explanation of the specific implementation of each step.

[0054] Figure 2The figure shows an example of an originally captured grayscale image, which contains a large amount of background information. By identifying the target region, subsequent steps can focus on the target region, avoiding redundant calculations across the entire image. This speeds up subsequent processing steps, meets real-time requirements, eliminates interference from irrelevant information, and improves the accuracy of locating the centroid of the main spatial target. Therefore, in step S1 of this embodiment, the target region containing the spatial target is first extracted from the originally captured grayscale image.

[0055] During the imaging process, due to the interference of random noise such as Gaussian noise and white noise, bright stray light spots or blurred areas may appear in the image, making the edge of the target unclear and the details difficult to identify. In addition, there are a large number of bright star points in the space environment. These noises and blurs not only affect the image quality, but also bring great challenges to the subsequent centroid positioning. Therefore, it is necessary to perform preprocessing to improve the image quality and the robustness of the subsequent algorithm. Based on these considerations, in step S1 of this embodiment, before extracting the grayscale image of the target area from the original collected grayscale image, preprocessing will be performed first. During the preprocessing process, the bright star points and noise are preliminarily suppressed by the denoising algorithm, and the technology of sharpening and enhancing the edges is used to make the target contour clearer, thereby providing a better basis for accurately and stably positioning the centroid of the subject.

[0056] Specifically, S1 of this embodiment includes:

[0057] S11: Grayscale image of the original collection I ori Gaussian blur processing is performed in sequence to reduce noise and star points; adaptive histogram equalization with limited contrast is performed on the Gaussian blurred image to enhance the contrast and detail effect of the target area to obtain the preprocessed image I pre ;

[0058] S12: Preprocessing images I pre Perform threshold adaptive binarization operation;

[0059] S13: Preprocessing image after binarization operation I pre Perform opening and closing operations to further suppress the highlight star noise and fill the holes in the binary area, making the target area smoother and more complete. Then perform dilation operations to connect adjacent target areas to obtain the image. I pre ' ;

[0060] S14: In the image Search for connected domains in the image and select the connected domain with the largest area as the target area.I pre Extract the target area and get the grayscale image of the target area I gray .

[0061] In spatial imaging, the complexity of the environment has a significant impact on the detection and positioning of the target. In addition, there are a large number of bright star points in the spatial environment, which can easily interfere with the subsequent target positioning steps. Since these star points have similar brightness values to the target area, they may be mistakenly identified as the characteristics 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 by the denoising algorithm, it is impossible to suppress the star points with similar brightness values to the target area. Therefore, step S2 of this embodiment further proposes a star point noise suppression method that can effectively suppress the star points in the target area based on the shape characteristics of the contours of these star point noises and the area of the enclosed area, which specifically includes:

[0062] Extract each contour in the target area and calculate the area of the area enclosed by each contour; the contour extraction in the target area can be performed on the grayscale image of the target area. I gray It is realized on the basis of threshold adaptive binarization operation;

[0063] Typically, stars are far away from space targets, and therefore, their area within the target region is relatively small. Based on this, in this embodiment, the areas enclosed by the contours, if their area is smaller than a preset area threshold, are determined as star points and filtered out.

[0064] This embodiment further considers that stars, due to their luminescence, form circular or approximately circular spots in the image. Therefore, the method for suppressing star points proposed in this embodiment further includes: determining, within the areas enclosed by each contour, areas with an area smaller than a preset area threshold as star points, filtering them out, calculating the minimum circumscribed circle of each remaining area, and calculating the ratio of the area of the area to the area of the minimum circumscribed circle as the fill rate of the corresponding area;

[0065] In the remaining area, the areas whose filling rate is greater than a preset filling rate threshold and whose minimum circumscribed circle radius is less than a preset roundness threshold are determined as star points and filtered out.

[0066] It is easy to understand that when suppressing star points in the target area, the relevant area threshold, fill rate threshold and roundness threshold can be determined accordingly based on the environmental characteristics of the current space detection scene.

[0067] In the processing of space targets with complex postures, the shape of the target may change significantly due to perspective effects. In order to improve the accuracy of recognition, it is usually necessary to normalize the target, that is, to perform standardized adjustment on the posture of the target image, so as to facilitate the identification and comparison of different areas. Based on this, this embodiment, through step S3, will first perform image plane normalization on the image before determining the main area of the space target. Specifically, in S3, the grayscale image of the target area after suppressing star point noise is normalized. I gray and binary images I bin The target areas in the image are all rotated to the level of the solar wing frame, including:

[0068] Extract the solar wing frame of the space target and calculate the angle between the extracted solar wing frame and the horizontal direction ；

[0069] According to the angle The grayscale image of the target area after suppressing star noise is I gray and binary images I bin The target area in the image is rotated until the solar wing frame in the target area is level.

[0070] Extract the solar wing skeleton of the space target, including:

[0071] Binary image after suppressing star noise I bin Perform straight line detection on the line segment and select the longest line segment from the detected line segments L 1 ;

[0072] Remove and line segments L 1 After finding the other collinear lines, find the longest line segment among the remaining line segments L 2 ;

[0073] Binary image after suppressing star noise I bin In the line segment L 1 and line segments L 2 The average slope of the slope is the slope, and a straight line is drawn through the centroid of the target area. L 3 , as the solar wing skeleton.

[0074] Different regions of the target exhibit significant differences in both the grayscale image and the binary image. Step S4 of this embodiment leverages these differences to effectively locate the main region of the spatial target. To accurately measure the differences between different regions within the spatial target, this embodiment comprehensively calculates the grayscale and binary differences between regions using both the grayscale and binary images. Optionally, this embodiment uses the normalized mean absolute error (MAE) as a measure of grayscale value differences. A lower MAE value indicates greater similarity between two regions. The calculation formula is as follows:

[0075] ;

[0076] in, G 1 and G 2 represents two areas in the grayscale image. and There are two areas in The gray value of a pixel, M is the total number of pixels in the region.

[0077] Optionally, in this embodiment, the normalized overlap coefficient (OC) is used as a measure of binary difference. OC measures similarity by counting the proportion of pixels with a value of 255 (i.e., white) at the same location in two regions. A higher OC value indicates more similarity between the two regions. The calculation formula is as follows:

[0078]

[0079] in, B 1 and B 2 represents two regions in the binary image; and There are two areas in The binary pixel value of each pixel (0 or 255), M is the total number of pixels in the region.

[0080] For each region z, the mean absolute error and overlap coefficient between the region and each other region are calculated respectively. The obtained mean absolute error and overlap coefficient are normalized respectively so that they can be compared at the same scale. Then the normalized mean absolute errors are added together to obtain D g , add the normalized overlap coefficients and get 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 =-0.5D g +0.5 D b , this comprehensive similarity measure is used to evaluate the overall similarity between region z and the remaining regions. It should be noted that when calculating the comprehensive similarity measure for a region, the weight setting can be set accordingly based on the actual spatial detection scenario. In certain specific cases, one of the weights can be set to 0, but both weights cannot be set to 0 at the same time.

[0081] After calculating the comprehensive similarity of each region, this embodiment determines the region with the smallest comprehensive similarity as the main region of the space target based on the characteristic that the main region has the greatest difference from the remaining regions, thereby accurately locating the main region in the space target.

[0082] For space targets of certain specific shapes, the centroid of the main body area may not be located on the space target. In response to this situation, this embodiment will further perform judgment and correction to ensure that the located main body centroid is located on the target.

[0083] against Figure 2 The grayscale image shown in the figure, the final centroid positioning result of the space target body obtained in this embodiment is as follows Figure 3 shown.

[0084] This embodiment further uses a typical two-wing target Deepspace for error testing. The imaging distance is set to 10 kilometers, and the average error in each posture is within 2 pixels. The error analysis results show that this embodiment can also achieve accurate and stable main body centroid positioning for space targets with diverse shapes and changing postures in complex environments.

[0085] Example 2:

[0086] A computer program product includes a computer program; when the computer program is executed by a processor, the method for locating the centroid of a space target provided in the above embodiment 1 is implemented.

[0087] Example 3:

[0088] A computer-readable storage medium includes a stored computer program; when the computer program is executed by a processor, the device where the computer-readable storage medium is located is controlled to execute the method for locating the centroid of a space target provided in the above-mentioned embodiment 1.

[0089] Example 4:

[0090] A space target body centroid positioning system, comprising:

[0091] a computer-readable storage medium for storing a computer program;

[0092] and a processor for reading a computer program stored in a computer-readable storage medium to implement the method for locating the centroid of a space target body provided in the above-mentioned embodiment 1.

[0093] It will be easily understood by those skilled in the art that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for locating the centroid of a space target, characterized in that: include: S1: Grayscale image collected from the original I ori Extract the target area containing the spatial target and obtain the grayscale image of the target area I gray ; S2: convert the grayscale image of the target area I gray Binarize to get a binary image I bin , and suppress the grayscale image of the target area respectively I gray and the binary image I bin Star point noise in S3: Grayscale image of the target area after suppressing star point noise I gray and binary images I bin The target areas in the image are rotated to the level of the solar wing skeleton to obtain the grayscale image with normalized image plane posture. I g_rot and binary images I b_rot ; S4: Divide the target area along the direction of the solar wing frame into N regions, for each region z, according to the grayscale map 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 map 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 object; 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, use 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 will be I bin The center of mass of the space target is used as the main centroid of the space target P .

2. The method for locating the centroid of a space target according to claim 1, wherein: In S2, star point noise suppression methods include: Extract each contour in the target area and calculate the area of the region enclosed by each contour; In the areas enclosed by each contour, the areas whose area is smaller than the preset area threshold are determined as star points and filtered out.

3. The method for locating the centroid of a space target according to claim 2, wherein: In S2, the method of suppressing star points further includes: determining, within the areas enclosed by the contours, areas whose areas are smaller than a preset area threshold as star points, filtering them out, calculating the minimum circumscribed circle of each remaining area, and calculating the ratio of the area of the area to the area of the minimum circumscribed circle as the fill rate of the corresponding area; In the remaining area, the areas whose filling rate is greater than a preset filling rate threshold and whose minimum circumscribed circle radius is less than a preset roundness threshold are determined as star points and filtered out.

4. The method for locating the centroid of a space target according to any one of claims 1 to 3, wherein: In S3, the grayscale image of the target area after suppressing the star point noise is I gray and binary images I bin The target areas in the image are all rotated to the level of the solar wing frame, including: Extract the solar wing frame of the space target and calculate the angle between the extracted solar wing frame and the horizontal direction ； According to the angle The grayscale image of the target area after suppressing star noise is I gray and binary images I bin The target area in the image is rotated until the solar wing frame in the target area is level.

5. The method for locating the centroid of a space target according to claim 4, wherein: Extract the solar wing skeleton of the space target, including: Binary image after suppressing star noise I bin Perform straight line detection on the line segment and select the longest line segment from the detected line segments L 1 ; Remove and line segments L 1 After finding the other collinear lines, find the longest line segment among the remaining line segments L 2 ; Binary image after suppressing star noise I bin In the line segment L 1 and line segments L 2 The average slope of the slope is the slope, and a straight line is drawn through the centroid of the target area. L 3 , as the solar wing skeleton.

6. The method for locating the centroid of a space target according to any one of claims 1 to 3, wherein: S1 includes: S11: the grayscale image originally collected I ori Gaussian blur processing and adaptive histogram equalization with limited contrast are performed in sequence to obtain the preprocessed image I pre ; S12: Preprocessing the image I pre Perform threshold adaptive binarization operation; S13: Preprocessing image after binarization operation I pre After performing the opening and closing operations, the expansion operation is performed to obtain the image I pre ' ; S14: In the image Search for connected domains in the image and select the connected domain with the largest area as the target area. I pre Extract the target area and get the grayscale image of the target area I gray .

7. The method for locating the centroid of a space target according to any one of claims 1 to 3, wherein: In S4, the grayscale value difference is measured by the normalized mean absolute error, and the binary value difference is measured by the normalized overlap coefficient.

8. A computer program product, characterized in that The method comprises a computer program; when the computer program is executed by a processor, the method for locating the centroid of a space target body according to any one of claims 1 to 7 is implemented.

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 method for locating the centroid of a space target body according to any one of claims 1 to 7.

10. A space target body centroid positioning system, characterized in that: include: a computer-readable storage medium for storing a computer program; and a processor, configured to read the computer program stored in the computer-readable storage medium to implement the method for locating the centroid of a space target body according to any one of claims 1 to 7.

Images