A geostationary orbit target detection method based on minimum pixel stacking

Abstract

The present invention discloses a method for detecting geostationary orbit targets based on minimum pixel stacking, which addresses the problem of how to accurately and quickly detect geostationary orbit targets. The method belongs to the field of space target detection and comprises the following steps: stacking the minimum pixel values ​​at the same pixel position in each frame of an image captured by a telescope in staring mode to generate a stacked image; calculating the pixel mean and standard deviation of the stacked image and setting a threshold based on the Laida criterion; performing threshold segmentation on the stacked image according to the threshold; labeling the connected domains using a connected domain labeling algorithm to obtain a binary image with outlier noise removed; searching for targets in the marked connected domains in the binary image using the connected domain labeling algorithm; and calculating the pixel coordinates and pixel values ​​of the targets using a Gaussian centroid positioning method to obtain the centroid coordinates of each target, thereby determining the position of the geostationary orbit target. The present invention amplifies the apparent motion difference between geostationary orbit targets and stars, enabling accurate and rapid detection of geostationary orbit targets.

Description

Technical Field

[0001] The present invention belongs to the technical field of space target detection and relates to a geostationary orbit target detection method based on minimum pixel superposition. Background Art

[0002] The geostationary orbit has zero orbital eccentricity and inclination, and the period of motion of space objects in this orbit is equal to the Earth's rotation period. Because the sub-satellite point trajectory of objects in geostationary orbit is a single point, high-value communications and remote sensing satellites often use geostationary orbit. Therefore, detection and monitoring of geostationary orbital objects are necessary to ensure operational safety.

[0003] Currently, detection of geostationary orbit targets primarily relies on large-field-of-view optical telescopes, which are low-cost and highly efficient, typically observing in staring mode. However, large-field-of-view telescopes present challenges such as dim target images, a large number of stars in the field of view, and significant interference from noise and stray light, making detection difficult. A commonly used detection method is stacking. By stacking multiple images, the signal of the geostationary orbit target in the stacked image is enhanced, while the noise, due to random errors, does not precisely add up. This significantly improves the target's signal-to-noise ratio and increases the detection success rate.

[0004] Although the stacking method can effectively improve the signal-to-noise ratio of the target, the large number of star images in the field of view will form tails after stacking, causing obstruction of the target. In addition, when the image is affected by stray light and presents a non-uniform background, the detection effect after stacking is also poor. In addition, it is necessary to remove the strip-like star images after stacking in order to detect the point-like geostationary orbit targets. Summary of the Invention

[0005] In order to solve the technical problem of how to remove the superimposed strip-shaped star images in images taken by a telescope in staring mode so as to accurately and quickly detect point-like geostationary orbit targets in the image, the present invention provides a geostationary orbit target detection method based on minimum pixel superposition. This method amplifies the apparent motion difference between geostationary orbit targets and stars through minimum pixel superposition, realizes the detection of geostationary orbit targets, effectively copes with the occlusion of strip-shaped stars, has a high detection rate and a fast detection speed for images under different conditions, can quickly and accurately detect and locate geostationary orbit targets in the telescope's field of view, and ensures the safe operation of geostationary orbit satellites.

[0006] The purpose of the present invention is specifically achieved through the following technical solutions:

[0007] The present invention discloses a method for detecting geostationary orbit targets based on minimum pixel superposition, the method comprising:

[0008] Step 1: Receive all images taken by the telescope in the staring mode, and superimpose the minimum pixel value at the same pixel position of each frame of the image to generate a superimposed image;

[0009] Step 2: Calculate the pixel mean and standard deviation of the stacked image respectively. Based on the Laida criterion, set a threshold according to the pixel mean and standard deviation of the stacked image. Perform threshold segmentation on the stacked image according to the threshold to obtain a segmented image. Use the connected component labeling algorithm to label the connected components of the segmented image to obtain a binary image with outlier noise removed.

[0010] Step 3: Use the connected domain labeling algorithm to find the targets in the marked connected domain in the binary image, count the number of targets and number them, and record the pixel coordinates of each target; use the Gaussian centroid positioning method to calculate the pixel coordinates and pixel values ​​of each target to obtain the centroid coordinates of each target, thereby obtaining the positions of all geostationary orbit targets in all images.

[0011] In step 1, the method of superimposing the minimum pixel value at the same pixel position of each frame image to generate a superimposed image includes:

[0012]

[0013] Among them, I min (x, y) is the pixel value of the superimposed image at the same pixel position, N is the total number of images taken by the telescope, t is the index of the image frame, t is 1 to N; I t (x,y) is the minimum pixel value at the same pixel position in each frame image, and (x,y) is the pixel position.

[0014] In step 2, the pixel mean of the superimposed image is calculated as follows:

[0015]

[0016] Wherein, μ is the pixel mean of the superimposed image, m×n is the size of the superimposed image, m is the maximum number of pixels of the superimposed image in the x-axis direction, and n is the maximum number of pixels of the superimposed image in the y-axis direction.

[0017] In step 2, the standard deviation of the superimposed image is calculated as:

[0018]

[0019] Wherein, σ is the standard deviation of the superimposed image, m×n is the size of the superimposed image, m is the maximum number of pixels of the superimposed image in the x-axis direction, and n is the maximum number of pixels of the superimposed image in the y-axis direction.

[0020] In step 2, the method of setting the threshold based on the Laida criterion and the pixel mean and standard deviation of the superimposed image includes:

[0021] threshold=μ+3σ;

[0022] Among them, threshold is the threshold, μ is the pixel mean of the superimposed image, and σ is the standard deviation of the superimposed image.

[0023] In step 2, the method of performing threshold segmentation on the superimposed image according to the threshold to obtain the segmented image includes:

[0024]

[0025] Among them, I binary (x,y) is the segmented image, I min (x, y) is the pixel value of the superimposed image at the same pixel position, and threshold is the threshold.

[0026] In step 2, a method for labeling connected components of the segmented image using a connected component labeling algorithm to obtain a binary image with outlier noise removed includes:

[0027] Use the connected domain marking algorithm to traverse all pixels row by row from the upper left corner of the segmented image; if the pixel value of the current pixel is 1, check the eight neighboring pixels of the current pixel. If no neighboring pixels are marked as connected domains among the eight neighboring pixels, create a connected domain for the current pixel and use the current pixel as the target;

[0028] If there is a target with a marked connected domain among the eight neighboring pixels, the current pixel with a pixel value of 1 is marked as part of the connected domain of the target;

[0029] After completing the connected domain labeling for all pixels in the segmented image using the connected domain labeling algorithm, the wild value noise with a connected domain area of ​​1 is removed to generate a binary image.

[0030] Among them, the connected domain labeling algorithm is used to identify and mark connected regions in the image, which is completed through the bwconncomp function in MATLAB.

[0031] In step 3, the pixel coordinates and pixel values ​​of the target are calculated using the Gaussian centroid positioning method. The method for obtaining the centroid coordinates of each target includes:

[0032]

[0033] Among them, (C x ,C y ) is the center of mass coordinate of the target, (x i ,y i ) is the i-th pixel coordinate in the target connected area, p i The pixel coordinate (x i ,yi )’s pixel value; i is the pixel coordinate number of the target connected area.

[0034] The beneficial effects of the present invention are:

[0035] The present invention adopts a minimum pixel superposition method. By superimposing the minimum pixel values ​​at the same pixel position of multiple frames of images taken by the telescope in staring mode, it effectively amplifies the apparent motion difference between stars and geostationary orbit targets, and can also handle the situation where stars block targets. Because geostationary orbit targets are stationary in the image, while stars have a speed of motion in the image, they show obvious tracks in consecutive multiple frames of images. Through the minimum pixel superposition method, each pixel in the multiple frames of images can be screened for the minimum value, thereby retaining the information of the geostationary orbit target and eliminating the motion track of the star, significantly improving the detection effect of the geostationary orbit target; including:

[0036] 1. Improve the detection rate of geostationary orbit targets:

[0037] Through the minimum pixel superposition method, the apparent motion difference between geostationary orbit targets and stars can be effectively amplified; in multi-frame images, the motion trajectory of the stars is eliminated. Compared with the traditional superposition method, the present invention has a significantly reduced missed detection rate in target detection, a higher detection rate, and more reliable detection results.

[0038] 2. Avoid star occlusion problems:

[0039] Star occlusion is a common problem in geostationary orbit target detection. Traditional overlay methods can obscure targets when they lie along star paths. The minimum pixel overlay method eliminates star paths by selecting the minimum pixel value for overlay, ensuring the target remains clearly visible in the final overlay image.

[0040] 3. Improve the anti-interference ability against abnormal bright spots:

[0041] Traditional overlay methods are susceptible to interference from unusual bright spots in some frames, leading to skewed target detection results. The minimum pixel overlay method, by filtering the pixel values ​​within each frame for the minimum value, suppresses the influence of these unusual bright spots and improves the algorithm's anti-interference capabilities. This method can still reliably detect targets even in the presence of unusual bright spots.

[0042] 4. Improve detection efficiency:

[0043] While maintaining high accuracy, the present invention improves detection efficiency by optimizing the algorithm structure. Compared with the traditional superposition method, the subsequent detection and positioning method of the present invention is simpler and significantly improves detection efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] The present invention will be described in further detail below with reference to the accompanying drawings and examples.

[0045] Figure 1 It is an original image frame received by the present invention from all images taken by the telescope in the staring mode.

[0046] Figure 2 In the present invention, the minimum pixel values ​​at the same pixel position of each frame image are superimposed to generate a superimposed image. DETAILED DESCRIPTION

[0047] An embodiment of the present invention provides a method for detecting geostationary orbit targets based on minimum pixel superposition, the method comprising:

[0048] Step 1: Receive all images taken by the telescope in staring mode, and superimpose the minimum pixel values ​​at the same pixel position of each frame of the image to generate a superimposed image; through the minimum pixel superposition method disclosed in the present invention, a new superimposed image can be generated, which retains the minimum pixel information in multiple frames of images at each pixel position.

[0049] Step 2: Calculate the pixel mean and standard deviation of the stacked image respectively. Based on the Laida criterion, set a threshold according to the pixel mean and standard deviation of the stacked image. Perform threshold segmentation on the stacked image according to the threshold to obtain a segmented image. Use the connected component labeling algorithm to label the connected components of the segmented image to obtain a binary image with outlier noise removed.

[0050] Step 3: Use the connected domain labeling algorithm to find the targets in the marked connected domain in the binary image, count the number of targets and number them, and record the pixel coordinates of each target; use the Gaussian centroid positioning method to calculate the pixel coordinates and pixel values ​​of each target to obtain the centroid coordinates of each target, thereby obtaining the positions of all geostationary orbit targets in all images.

[0051] In step 1, the method of superimposing the minimum pixel value at the same pixel position of each frame image to generate a superimposed image includes:

[0052]

[0053] Among them, I min (x, y) is the pixel value of the superimposed image at the same pixel position, N is the total number of images taken by the telescope, t is the index of the image frame, t is 1 to N; I t (x,y) is the minimum pixel value at the same pixel position in each frame image, and (x,y) is the pixel position.

[0054] In step 2, the pixel mean is the sum of all pixel values ​​in the image divided by the total number of pixels. The pixel mean of the superimposed image is calculated as:

[0055]

[0056] Wherein, μ is the pixel mean of the superimposed image, m×n is the size of the superimposed image, m is the maximum number of pixels of the superimposed image in the x-axis direction, and n is the maximum number of pixels of the superimposed image in the y-axis direction.

[0057] In step 2, the pixel standard deviation is the square root of the mean of the squares of the differences between all pixel values ​​in the image and the mean. The standard deviation of the stacked image is calculated as:

[0058]

[0059] Wherein, σ is the standard deviation of the superimposed image, m×n is the size of the superimposed image, m is the maximum number of pixels of the superimposed image in the x-axis direction, and n is the maximum number of pixels of the superimposed image in the y-axis direction.

[0060] In step 2, the method of setting the threshold based on the Laida criterion and the pixel mean and standard deviation of the superimposed image includes:

[0061] threshold=μ+3σ;

[0062] Among them, threshold is the threshold, μ is the pixel mean of the superimposed image, and σ is the standard deviation of the superimposed image.

[0063] In step 2, the method of performing threshold segmentation on the superimposed image according to the threshold to obtain the segmented image includes:

[0064]

[0065] Among them, I binary (x,y) is the segmented image, I min (x, y) is the pixel value of the superimposed image at the same pixel position, and threshold is the threshold.

[0066] In step 2, a method for labeling connected components of the segmented image using a connected component labeling algorithm to obtain a binary image with outlier noise removed includes:

[0067] Use the connected domain marking algorithm to traverse all pixels row by row from the upper left corner of the segmented image; if the pixel value of the current pixel is 1, check the eight neighboring pixels of the current pixel. If no neighboring pixels are marked as connected domains among the eight neighboring pixels, create a connected domain for the current pixel and use the current pixel as the target;

[0068] If there is a target with a marked connected domain among the eight neighboring pixels, the current pixel with a pixel value of 1 is marked as part of the connected domain of the target;

[0069] After completing the connected domain labeling for all pixels in the segmented image using the connected domain labeling algorithm, the wild value noise with a connected domain area of ​​1 is removed to generate a binary image.

[0070] Among them, the connected domain labeling algorithm is mainly used to identify and mark connected areas in the image, which is completed through the bwconncomp function in MATLAB.

[0071] In step 3, the pixel coordinates and pixel values ​​of the target are calculated using the Gaussian centroid positioning method: the coordinates of each pixel of the target are multiplied by its corresponding pixel value, and then all weighted coordinates are summed and divided by the sum of the target pixel values. The method for obtaining the centroid coordinates of each target includes:

[0072]

[0073] Among them, (C x ,C y ) is the center of mass coordinate of the target, (x i ,y i ) is the i-th pixel coordinate in the target connected area, p i The pixel coordinate (x i ,y i ) pixel value; i is the pixel coordinate number of the target connected area. This method ultimately determines the location of all geostationary orbit targets in the image.

[0074] Verification Example

[0075] Based on the motion characteristics of geostationary orbit satellites and stars, since geostationary orbit targets are stationary relative to the Earth, when the telescope is working in staring mode, they are stationary points in the telescope's field of view. The angular velocity of the star in the telescope's field of view is related to its declination. The specific formula is:

[0076] |ω|=Ω·cosθ

[0077] Where Ω is the angular velocity of the Earth's rotation, and θ is the declination of the star. That is, the angular velocity of the stars in the field of view is determined by the declination of the telescope, and its value range is [0, Ω]. When the telescope is pointed at the north and south celestial poles, the angular velocity of the stars in the field of view is 0; when the telescope is pointed at the celestial equator, the angular velocity of the stars in the field of view is Ω. The present invention can use the angular velocity ω of the target relative to the telescope and the angular resolution θ of the telescope to calculate the angular velocity of the star. P The correlation between them is used to obtain the target's motion speed v in the pixel coordinate system. T size,

[0078]

[0079] Among them, θ Pis the angular resolution of the telescope, which refers to the minimum angle that the telescope can resolve. The calculation formula is as follows:

[0080]

[0081] Among them, Θ FOV represents the field of view of the telescope, and R represents the image resolution. Therefore, the stars appear to be moving on the image.

[0082] Based on the above research, the present invention uses the minimum pixel superposition method to amplify the apparent motion difference between geostationary orbit targets and stars, thereby realizing the detection of geostationary orbit targets. The experimental steps are as follows:

[0083] Receive all images taken by the telescope in staring mode: the telescope works in staring mode, the exposure time is 2s, the telescope points to the declination of 0°, and the total number of frames N of the sequence image is 29.

[0084] The minimum pixel value at the same pixel position of each frame image is superimposed to generate a superimposed image: perform minimum pixel superposition on all 29 frames of images, where one frame of the original image is as follows: Figure 1 As shown, the superimposed image is generated as Figure 2 shown.

[0085] The pixel mean and standard deviation of the superimposed image were calculated respectively: the pixel mean μ was 863.2, and the standard deviation σ was 44.5;

[0086] According to the Laida criterion, the threshold is calculated to be 996.7;

[0087] Based on this threshold, the image is segmented, the noise points are removed after binarization, and the number of targets and the pixel position index corresponding to each target are recorded.

[0088] Target Detection and Localization: Detected targets are processed sequentially. Based on the target's pixel position index, the pixel values ​​at the corresponding positions are extracted from the overlay image. The target's center of mass is calculated using the Gaussian centroid localization method to complete the target location. The final number of detected geostationary satellites is 5, and the center of mass locations are (1552.0, 2146.6), (1598.5, 2236.2), (2208.5, 1960.6), (2899.9, 1776.5), and (3477.5, 1582.2).

[0089] The beneficial effects of the embodiments of the present invention are:

[0090] The present invention adopts a minimum pixel superposition method. By superimposing the minimum pixel values ​​at the same pixel position of multiple frames of images taken by the telescope in staring mode, it effectively amplifies the apparent motion difference between stars and geostationary orbit targets, and can also handle the situation where stars block targets. Because geostationary orbit targets are stationary in the image, while stars have a speed of motion in the image, they show obvious tracks in consecutive multiple frames of images. Through the minimum pixel superposition method, each pixel in the multiple frames of images can be screened for the minimum value, thereby retaining the information of the geostationary orbit target and eliminating the motion track of the star, significantly improving the detection effect of the geostationary orbit target; including:

[0091] 1. Improve the detection rate of geostationary orbit targets:

[0092] Through the minimum pixel superposition method, the apparent motion difference between geostationary orbit targets and stars can be effectively amplified; in multi-frame images, the motion trajectory of the stars is eliminated. Compared with the traditional superposition method, the present invention has a significantly reduced missed detection rate in target detection, a higher detection rate, and more reliable detection results.

[0093] 2. Avoid star occlusion problems:

[0094] Star occlusion is a common problem in geostationary orbit target detection. Traditional overlay methods can obscure targets when they lie along star paths. The minimum pixel overlay method eliminates star paths by selecting the minimum pixel value for overlay, ensuring the target remains clearly visible in the final overlay image.

[0095] 3. Improve the anti-interference ability against abnormal bright spots:

[0096] Traditional overlay methods are susceptible to interference from unusual bright spots in some frames, leading to skewed target detection results. The minimum pixel overlay method, by filtering the pixel values ​​within each frame for the minimum value, suppresses the influence of these unusual bright spots and improves the algorithm's anti-interference capabilities. This method can still reliably detect targets even in the presence of unusual bright spots.

[0097] 4. Improve detection efficiency:

[0098] While maintaining high accuracy, the present invention improves detection efficiency by optimizing the algorithm structure. Compared with the traditional superposition method, the subsequent detection and positioning method of the present invention is simpler and significantly improves detection efficiency.

[0099] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

Claims

1. A method for detecting geostationary orbit targets based on minimum pixel superposition, characterized in that: The method includes: Step 1: Receive all images taken by the telescope in the staring mode, and superimpose the minimum pixel value at the same pixel position of each frame of the image to generate a superimposed image; Step 2: Calculate the pixel mean and standard deviation of the stacked image respectively. Based on the Laida criterion, set a threshold according to the pixel mean and standard deviation of the stacked image. Perform threshold segmentation on the stacked image according to the threshold to obtain a segmented image. Use the connected component labeling algorithm to label the connected components of the segmented image to obtain a binary image with outlier noise removed. Step 3: Use the connected domain labeling algorithm to find the targets in the marked connected domain in the binary image, count the targets and number them, and record the pixel coordinates of each target; use the Gaussian centroid positioning method to calculate the pixel coordinates and pixel values ​​of each target to obtain the centroid coordinates of each target, thereby obtaining the positions of all geostationary orbit targets in all images; In step 1, the method of superimposing the minimum pixel value at the same pixel position of each frame image to generate a superimposed image includes: Among them, I min (x, y) is the pixel value of the superimposed image at the same pixel position, N is the total number of images taken by the telescope, t is the index of the image frame, t is 1 to N; I t (x,y) is the minimum pixel value at the same pixel position in each frame image, and (x,y) is the pixel position.

2. The method according to claim 1, wherein In step 2, the pixel mean of the superimposed image is calculated as follows: Wherein, μ is the pixel mean of the superimposed image, m×n is the size of the superimposed image, m is the maximum number of pixels of the superimposed image in the x-axis direction, and n is the maximum number of pixels of the superimposed image in the y-axis direction.

3. The method according to claim 1, wherein In step 2, the standard deviation of the superimposed image is calculated as: Wherein, σ is the standard deviation of the superimposed image, m×n is the size of the superimposed image, m is the maximum number of pixels of the superimposed image in the x-axis direction, and n is the maximum number of pixels of the superimposed image in the y-axis direction.

4. The method according to any one of claims 1 to 3, characterized in that In step 2, the method of setting the threshold based on the Laida criterion and the pixel mean and standard deviation of the superimposed image includes: threshold=μ+3σ; Among them, threshold is the threshold, μ is the pixel mean of the superimposed image, and σ is the standard deviation of the superimposed image.

5. The method according to claim 4, wherein In step 2, the method of performing threshold segmentation on the superimposed image according to the threshold to obtain the segmented image includes: Among them, I binary (x,y) is the segmented image, I min (x, y) is the pixel value of the superimposed image at the same pixel position, and threshold is the threshold.

6. The method according to claim 5, wherein In step 2, a method for labeling connected components of the segmented image using a connected component labeling algorithm to obtain a binary image with outlier noise removed includes: Use the connected domain marking algorithm to traverse all pixels row by row from the upper left corner of the segmented image; if the pixel value of the current pixel is 1, check the eight neighboring pixels of the current pixel. If no neighboring pixels are marked as connected domains among the eight neighboring pixels, create a connected domain for the current pixel and use the current pixel as the target; If there is a target with a marked connected domain among the eight neighboring pixels, the current pixel with a pixel value of 1 is marked as part of the connected domain of the target; After completing the connected domain labeling for all pixels in the segmented image using the connected domain labeling algorithm, the wild value noise with a connected domain area of ​​1 is removed to generate a binary image.

7. The method according to claim 1 or 6, wherein: The connected region marking algorithm is used to identify and mark connected regions in an image and is implemented using the bwconncomp function in MATLAB.

8. The method according to claim 1, wherein In step 3, the pixel coordinates and pixel values ​​of the target are calculated using the Gaussian centroid positioning method. The method for obtaining the centroid coordinates of each target includes: Among them, (C x ,C y ) is the center of mass coordinate of the target, (x i ,y i ) is the i-th pixel coordinate in the target connected area, p i The pixel coordinate (x i ,y i )’s pixel value; i is the pixel coordinate number of the target connected area.

Images