Star map segmentation method based on local background estimation
By adopting a star map segmentation method based on local background estimation in the star sensor, the problem of difficulty in extracting star targets under complex backgrounds is solved, and accurate extraction and robust adaptation under different background conditions are achieved, providing reliable support for the attitude measurement of the star sensor.
Patent Information
- Application Number
- CN202510104919.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-01-23
- Publication Date
- 2025-05-23
AI Technical Summary
The prior art cannot efficiently and accurately extract stellar targets in complex contexts, resulting in increased difficulty in measuring postures of star sensors.
Using the star map segmentation method based on local background estimation, an effective segmentation strategy is designed to distinguish background from targets through steps such as image denoising, calculating local gradients, local median and standard deviations, adaptive threshold calculations and image threshold segmentation.
Accurately extracting star dot targets under different background conditions, showing good adaptability and robustness, providing reliable technical support for the attitude measurement of star sensors.
Smart Images
Figure CN120031907A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of star sensors, and particularly to a star map segmentation method based on local background estimation. Background Art
[0002] A star sensor is an optical device used for spacecraft attitude measurement, which can determine its own attitude by capturing starlight in the starry sky. The image processing technology of star sensors is one of the core technologies of this device, involving processing the captured starry sky image, extracting the centroid of stars from it, so as to realize star map recognition and attitude calculation.
[0003] In the working environment of star sensors, they will be affected by celestial bodies such as the sun and the moon, as well as stray light in low earth orbit, increasing the background brightness of the image, resulting in a decrease in the signal-to-noise ratio of star targets and increasing the difficulty of extracting star targets from them.
[0004] In summary, it is crucial to find a method that can efficiently and accurately extract star targets in complex backgrounds. Summary of the Invention
[0005] The present invention aims to solve the technical problem in the prior art that star targets cannot be efficiently and accurately extracted in complex backgrounds. In view of the different characteristics of the background region and the target region, an effective segmentation strategy is designed, and a star map segmentation method based on local background estimation is provided.
[0006] To solve the above technical problems, the technical solution of the present invention is specifically as follows:
[0007] A star map segmentation method based on local background estimation includes the following steps:
[0008] Step 1: Image denoising;
[0009] Use spatial domain filtering or frequency domain filtering methods to remove image noise and enhance the effective information of the image;
[0010] Step 2: Calculate local gradients;
[0011] For each pixel I(x, y) in the image I, extract a 5×5 region centered on this pixel, divide the 5×5 region into four quadrants, calculate the gradient of each quadrant, and then take the average value of the gradients of the four quadrants;
[0012] Step 3: Calculate local medians;
[0013] For each pixel I(x, y) in the image I, extract all the pixels in a 5×5 region centered on this pixel and take its local median;
[0014] Step 4: Calculate local standard deviations;
[0015] For each pixel I(x,y) in image I, extract all pixels in a 5×5 region centered on the pixel and calculate the local standard deviation;
[0016] Step 5: Calculate the adaptive threshold;
[0017] Calculate the local threshold based on the local median, local gradient mean and local standard deviation;
[0018] Step 6: Image threshold segmentation;
[0019] Perform threshold segmentation on the image based on the local threshold.
[0020] In the above technical solution, step one is specifically as follows:
[0021] Given a Gaussian kernel:
[0022]
[0023] For each pixel I(x,y) in image I, when calculating the new pixel value, it is necessary to extract a 3×3 area centered on the pixel, multiply it element-by-element with the Gaussian kernel in the above formula, sum it, and finally perform normalization:
[0024]
[0025] In the above technical solution, step 2 is specifically as follows:
[0026]
[0027] Step 2.1 Calculate the first quadrant gradient:
[0028]
[0029] Step 2.2 Calculate the second quadrant gradient:
[0030]
[0031] Step 2.3 Calculate the third quadrant gradient:
[0032]
[0033] Step 2.4 Calculate the fourth quadrant gradient:
[0034]
[0035] Step 2.5 Calculate the local gradient mean S mean (x,y):
[0036]
[0037] In the above technical solution, step three is specifically as follows:
[0038] Local median I mid (x,y) satisfies:
[0039]
[0040] Among them, mid[] represents the data median function.
[0041] In the above technical solution, step 4 is specifically as follows:
[0042] Step 4.1 Calculate the average value of all pixels in the 5×5 area:
[0043]
[0044] Step 4.2 Calculate the standard deviation of all pixels in the 5×5 area:
[0045]
[0046] In the above technical solution, step five is specifically as follows:
[0047] The local threshold T(x,y) is:
[0048]
[0049] Where a is the coefficient.
[0050] In the above technical solution, step six is specifically as follows:
[0051]
[0052] Among them, I bin (x, y) is the result of threshold segmentation of each pixel I(x, y) in image I, 1 represents foreground pixel and 0 represents background pixel.
[0053] The present invention has the following beneficial effects:
[0054] The star image segmentation method based on local background estimation of the present invention designs an effective segmentation strategy by analyzing the significant differences in gradient and standard deviation between the background area and the target area; utilizes the local grayscale median and adaptive threshold determination method of the image to accurately extract star point targets under different background conditions, showing good adaptability and robustness, and providing reliable technical support for the attitude measurement of star sensors. BRIEF DESCRIPTION OF THE DRAWINGS
[0055] The present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments.
[0056] Figure 1It is a schematic diagram of the flow of the star image segmentation method based on local background estimation of the present invention. DETAILED DESCRIPTION
[0057] The inventive concept of the present invention is:
[0058] Considering that the background area and the target area have significant differences in gradient and standard deviation, the method of the present invention can effectively distinguish the background from the target through these feature differences. Specifically, by obtaining the local grayscale median of the image and combining the local gradient and standard deviation features, the image threshold is dynamically determined. This adaptive threshold method can maintain accurate target extraction capabilities under different background conditions.
[0059] The star map segmentation method based on local background estimation of the present invention has good adaptability and robustness, can accurately extract star point targets from complex backgrounds, and provides reliable technical support for attitude measurement of star sensors.
[0060] The present invention is described in detail below with reference to the accompanying drawings.
[0061] like Figure 1 As shown, the star image segmentation method based on local background estimation of the present invention is implemented by the following steps:
[0062] Step 1: Image denoising.
[0063] Use spatial domain filtering or frequency domain filtering methods to remove image noise and enhance the effective information of the image.
[0064] Taking Gaussian filtering as an example, given the Gaussian kernel:
[0065]
[0066] For each pixel I(x, y) in the image I, when calculating the new pixel value, it is necessary to extract a 3×3 region centered on the pixel, perform element-by-element multiplication with the Gaussian kernel in formula (1), sum it, and finally perform normalization. The above “3×3 region” refers to: a 3 pixel × 3 pixel region.
[0067]
[0068] Step 2: For each pixel I(x,y) in image I, extract a 5×5 region centered on the pixel, divide the 5×5 region into four quadrants, calculate the gradient of each quadrant, and then take the average of the gradients of the four quadrants. The above “5×5 region” refers to: a 5 pixel × 5 pixel region.
[0069]
[0070] Step 2.1 Calculate the first quadrant gradient:
[0071]
[0072] Step 2.2 Calculate the second quadrant gradient:
[0073]
[0074] Step 2.3 Calculate the third quadrant gradient:
[0075]
[0076] Step 2.4 Calculate the fourth quadrant gradient:
[0077]
[0078] Step 2.5 Calculate the local gradient mean S mean (x,y):
[0079]
[0080] Step 3: For each pixel I(x,y) in image I, extract all pixels in a 5×5 area centered on the pixel and take their local median.
[0081] Local median I mid (x,y) satisfies:
[0082]
[0083] Among them, mid[] represents the data median function.
[0084] Step 4: For each pixel I(x,y) in image I, extract all pixels in a 5×5 region centered on the pixel and calculate the local standard deviation.
[0085] Step 4.1 Calculate the average value of all pixels in the 5×5 area:
[0086]
[0087] Step 4.2 Calculate the standard deviation of all pixels in the 5×5 area:
[0088]
[0089] Step 5: According to the local median I mid (x,y), local gradient mean S mean (x,y) and the local standard deviation I std (x,y) calculates the local threshold T(x,y).
[0090] The local threshold T(x,y) is:
[0091]
[0092] Here, a is a coefficient, and in this example a=3.
[0093] Step 6: Perform threshold segmentation on the image based on the local threshold:
[0094]
[0095] Among them, I bin (x, y) is the result of threshold segmentation of each pixel I(x, y) in image I, 1 represents foreground pixel and 0 represents background pixel.
[0096] The star image segmentation method based on local background estimation of the present invention designs an effective segmentation strategy by analyzing the significant differences in gradient and standard deviation between the background area and the target area; utilizes the local grayscale median and adaptive threshold determination method of the image to accurately extract star point targets under different background conditions, showing good adaptability and robustness, and providing reliable technical support for the attitude measurement of star sensors.
[0097] Obviously, the above embodiments are merely examples for the purpose of clear explanation, and are not intended to limit the implementation methods. For those skilled in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to list all the implementation methods here. The obvious changes or modifications derived therefrom are still within the scope of protection of the invention.
Claims
1. A star image segmentation method based on local background estimation, characterized in that: The following steps are involved: Step 1: Image denoising; Use spatial domain filtering or frequency domain filtering methods to remove image noise to enhance the effective information of the image; Step 2: Calculate the local gradient; For each pixel I(x,y) in image I, extract a 5×5 region centered on the pixel, divide the 5×5 region into four quadrants, calculate the gradient of each quadrant, and then take the average of the gradients of the four quadrants; Step 3: Calculate the local median; For each pixel I(x,y) in image I, extract all pixels in a 5×5 region centered on the pixel and take their local median; Step 4: Calculate the local standard deviation; For each pixel I(x,y) in image I, extract all pixels in a 5×5 region centered on the pixel and calculate the local standard deviation; Step 5: Calculate the adaptive threshold; Calculate the local threshold based on the local median, local gradient mean and local standard deviation; Step 6: Image threshold segmentation; Perform threshold segmentation on the image based on the local threshold.
2. The star image segmentation method based on local background estimation according to claim 1, characterized in that: Step 1 is as follows: Given a Gaussian kernel: For each pixel I(x,y) in image I, when calculating the new pixel value, it is necessary to extract a 3×3 area centered on the pixel, multiply it element-by-element with the Gaussian kernel in the above formula, sum it, and finally perform normalization:
3. The star image segmentation method based on local background estimation according to claim 1, characterized in that: Step 2 is as follows: Step 2.1 Calculate the first quadrant gradient: Step 2.2 Calculate the second quadrant gradient: Step 2.3 Calculate the third quadrant gradient: Step 2.4 Calculate the fourth quadrant gradient: Step 2.5 Calculate the local gradient mean S mean (x,y):
4. The star image segmentation method based on local background estimation according to claim 3, characterized in that: Step three is as follows: Local median I mid (x,y) satisfies: Among them, mid[] represents the data median function.
5. The star image segmentation method based on local background estimation according to claim 4, characterized in that: Step 4 is as follows: Step 4.1 Calculate the average value of all pixels in the 5×5 area: Step 4.2 Calculate the standard deviation of all pixels in the 5×5 area:
6. The star image segmentation method based on local background estimation according to claim 5, characterized in that: Step 5 is as follows: The local threshold T(x,y) is: Where a is the coefficient.
7. The star image segmentation method based on local background estimation according to claim 1, characterized in that: Step 6 is as follows: Among them, I bin (x, y) is the result of threshold segmentation of each pixel I(x, y) in image I, 1 represents foreground pixel and 0 represents background pixel.