A method for calculating star trail masks in space target observation image processing

Abstract

A method for calculating star trail masks in space target observation image processing is presented, belonging to the field of space target observation image processing technology. It addresses the problems of low processing efficiency, false alarm rates, and insufficient detection rates in existing space target detection methods, whether frame difference or mask methods, which rely heavily on image recognition algorithms. First, morphological methods are used to determine the width of star trails at specific lengths and magnitudes. Second, statistical methods are used to determine the length, distortion width, and slope of the star trails in the image to be processed. Then, star map matching and high-magnitude star coordinate inversion are performed. Finally, star trail dilation and rotation transformations are applied to the inverted star information to form a star trail mask. This method can successfully mask star trails within a specified magnitude range in an image, significantly improving the detection efficiency and recognition rate of space targets while reducing the false alarm rate, and has broad application scenarios.

Description

Technical Field

[0001] This invention belongs to the field of space target observation image processing technology, and specifically relates to a method for calculating star trail masks for space target observation image processing. Background Technology

[0002] With the reduction in space launch costs and the rise of large satellite constellations, the number of spacecraft in space is constantly increasing. Therefore, the effective detection of space targets is crucial to ensuring the safety and sustainable development of space activities. Among these measures, using space-based satellites to determine the orbit of space targets is an important means of effectively monitoring them.

[0003] Because space-based observation payloads have powerful detection capabilities and a vast amount of stellar data in their field of view, and because space targets and stellar orbits have similar shapes and textures in images, the sheer number of stars without pre-removal can slow down the overall image processing speed and cause exponential growth in multi-frame correlation processing time, leading to the curse of dimensionality. Excessive targets in the candidate region can also hinder analysis efficiency when calculating track results across multiple frames. Therefore, it is necessary to remove stars from the image beforehand.

[0004] Currently, the processing of stars in space target detection data mainly employs image processing techniques, including frame differencing and masking methods. Frame differencing relies on processing multiple consecutive frames of images. Without precise star registration, additional noise is introduced; conversely, image registration increases the computational load during frame differencing. Even slight jitter on the satellite platform can lead to additional residual noise at star edges. Masking methods, in practice, often use fixed thresholds. While image filtering or morphological operations can remove residual noise, these operations can distort the target information in the original image. Both frame differencing and masking methods are image processing techniques, and their overall drawback lies in their reliance on the level of image recognition algorithms. Low processing efficiency, false alarm rates, and missed detection rates are all shortcomings of these methods. Summary of the Invention

[0005] To address the issues of low processing efficiency, false alarm rate, and insufficient false detection rate inherent in existing space target detection methods, both frame difference and mask methods, which rely heavily on image recognition algorithms, this invention provides a star orbit mask calculation method for space target observation image processing. The method specifically comprises: S1. The relationship between star orbit width, star orbit length, and star magnitude is determined by exponential fitting. S2. Provide the star trail length, distortion width, and slope of the image to be processed through sample statistics; S3. Use the identified low-magnitude stars to perform star map matching and invert the coordinates of high-magnitude stars; S4. Generate a star trail mask based on the star trail length, twist width, and star trail width; S5. Use the slope of star orbits to perform star orbit rotation transformations; S6, repeating steps S4 and S5, generates a rectangular mask for all star points, and merges the masks to create the final image mask for removing star trails.

[0006] Furthermore, the relationship between the width of the star orbit, the length of the star orbit, and the magnitude of the star is as follows: ;in, and Represents the fitted parameters, Indicates the width of the star trails. Indicates the length of the star orbit. This indicates the length of star trails in a standard star trail image. Indicates the magnitude of a star. This indicates that the star trail width is redundant.

[0007] Furthermore, step S2 specifically includes: For the image to be processed, the pixel coordinates of the four corner points and the center pixel coordinates of the star trails of low magnitude stars are obtained through image recognition algorithms. The length of each identified star trail is then calculated using the coordinates of the four corner points. and width ; Based on the identified star trail length The 10% truncation mean is used as the final star trail length. ; For the identified star trail width The mean of the 10% cutoff is used as the final twist width. ; For the star orbit slope, the mean value with a 10% cutoff is calculated as the final star orbit slope. .

[0008] Furthermore, the low-magnitude stars are those with relatively high brightness that can be extracted through image recognition and used for star map matching, while the high-magnitude stars are those with relatively low brightness that cannot be extracted or are extracted inaccurately through image recognition. The distinction between low-magnitude and high-magnitude stars is made based on specific circumstances.

[0009] Furthermore, step S3 specifically involves: using the star orbit center pixel coordinates obtained in step S2 to perform star map matching, and inverting the center pixel coordinates of high-magnitude stars based on the matching results and the high-magnitude star catalog. The range of star magnitudes selected for the high-magnitude star catalog is less than the threshold. threshold Star orbit length Linear functions:

[0010] The coefficients of the above linear function , By fitting the test data of the optical camera and taking star charts of stars with different orbit lengths, the maximum magnitude that needs to be masked is analyzed, and then the coefficients of the above formula are fitted.

[0011] Furthermore, in step S4, the coordinates of the four corner points of the star trail mask are calculated. They are respectively: ; ; ; .

[0012] Furthermore, step S5 specifically involves: selecting the rotation angle. , Rotation matrix for: Then the final coordinates of the four corner points for: ; ; ; .

[0013] The present invention also provides a star trail mask calculation system for space target observation image processing, the system comprising: The unit that calibrates the relationship between star orbit width, star orbit length, and star magnitude is determined by exponential fitting; The unit values ​​of star trail length, distortion width, and slope of the image to be processed are given through sample statistics. A unit that uses identified low-magnitude stars to perform star map matching and invert the coordinates of high-magnitude stars; The star trail mask is generated based on the star trail length, twist width, and star trail width. A unit that uses the slope of star orbits to perform star orbit rotation transformations; The process iterates to generate rectangular masks for all star points, and then merges these masks to create the final image mask for removing star trails.

[0014] The beneficial effects of the method described in this invention are as follows: First, the width of star trails at a specific length and magnitude is given through morphological methods. Second, the length, distortion width, and slope of the star trails in the image to be processed are given through sample statistics. Then, star map matching and high-magnitude star coordinate inversion are performed. Finally, star trail dilation and rotation transformations are applied to the inverted star point information to form a star trail mask. Using this method, star trails within a specified magnitude range in the image can be successfully masked, thereby preserving space target information. This can significantly improve the detection efficiency and recognition rate of space targets, reduce the false alarm rate of space targets, and has a wide range of application scenarios. Using the method described in this invention to calculate star trail masks for space target observation image data has the following advantages: (1) It can significantly improve the detection efficiency and recognition rate of space targets and reduce the false alarm rate of space targets; (2) The formula used uses the test fitting data parameters and does not depend on the specific parameters of the optical camera, so the method is universal; (3) The star trail mask obtained by using star catalog inversion is accurate and has little impact on target information in the image; (4) The size and number of masks can be dynamically adjusted according to the magnitude of the star, making it flexible in use; (5) The star trail width is calculated using the twist width parameter, which has a greater tolerance for platform jitter; (6) Using mathematical and geometric methods to invert the mask is more computationally efficient than image processing methods. Attached Figure Description

[0015] Figure 1 This is a schematic diagram showing the coordinate positions of the four corner points of the star-track rectangular mask in an embodiment of the present invention; Figure 2 This is a fitted curve of star magnitude and star orbit width in an embodiment of the present invention; Figure 3 This is a schematic diagram of the original image in an embodiment of the present invention; Figure 4 This is a schematic diagram of the processed image in an embodiment of the present invention. Detailed Implementation

[0016] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the protection scope of the present invention.

[0017] Example 1 The main objective of this invention is to provide a mathematical and geometric method for masking star trails. This method first calibrates the star trail width formula using an exponential fitting method. Then, it performs coarse identification on each frame of the image to be processed, statistically analyzing the length, width, and slope of the star trails. Next, it calculates the center position of each star trail using star map matching and inversion methods. Finally, it performs point-by-point calculations and merges all star trails that need to be masked to obtain the final star trails in the masked image.

[0018] The method is specifically as follows: Step 1: The relationship between the width of the star trail and the length of the star trail, as well as the magnitude of the star, is determined by exponential fitting.

[0019] When a satellite images a space target, the different energies of stars of different magnitudes will ultimately result in different widths of their orbits. Since star magnitudes are calculated using logarithmic coordinates, the width of the orbits varies. With stars and magnitude The relationship can be fitted using an exponential function: (1) A standard star trail image was selected for parameter calibration, and the magnitudes of different stars in the image were calculated manually. The width of the star trails below The parameters can then be fitted. , Simultaneously record the length of the star trails in this frame image. .

[0020] To eliminate fitting residuals and ensure that the final generated star trail mask can completely cover the star trails, the above formula needs to add a certain amount of star trail width redundancy. ,Right now: (2) in The selection should ensure that the width of all star trails used for fitting is less than [a certain value]. That is, you can usually directly select the fitted residual. Value and expand appropriately.

[0021] When a satellite images a space target, since the aperture and frame rate of the optical camera are fixed, the energy of the star trails captured on the focal plane for a star of a specific magnitude is also fixed. Therefore, when the magnitude is fixed, the width of the star trails is... With the length of star trails They are directly proportional, that is: (3) Combining equations (1), (2), and (3) yields the width of the star trails. With star orbit length Stars, etc. Relationship: (4).

[0022] Step two: Analyze the sample statistics to determine the length of the star trails, the width of the distortion, and the slope of the image to be processed.

[0023] For the image to be processed, image recognition algorithms can obtain the pixel coordinates of the four corner points and the center pixel coordinates of the star trails of low-magnitude stars. The length of each identified star trail can be calculated using the coordinates of the four corner points. and width Due to the influence of signal-to-noise ratio, the recognition rate of star trails of stars with excessively high magnitudes is low and the false alarm rate is high. Therefore, sample statistics are performed on the extracted star trails.

[0024] Based on the identified star trail length The 10% truncation mean is used as the final star trail length. ; Regarding the width of the star orbit, the perceived width can be affected by satellite platform jitter. It will eventually be larger than the actual width of the star trails. Similarly, the mean of the 10% cutoff is used as the final twist width. .

[0025] For the star orbit slope, the mean value with a 10% cutoff is also calculated as the final star orbit slope. .

[0026] Step 3: Use the identified low-magnitude stars to perform star map matching and invert the coordinates of high-magnitude stars.

[0027] Low-magnitude stars are relatively bright and easily extracted using image recognition methods, making them suitable for star map matching. High-magnitude stars are dimmer and difficult or inaccurate to identify, so they are extracted for inversion rather than image recognition. Specific distinction criteria may exist for particular optical systems, but this invention is not limited to specific optical systems; those skilled in the art can make distinctions based on specific circumstances.

[0028] Star map matching is performed using the center pixel coordinates of star orbits obtained in step two. The star map matching algorithm is relatively mature and will not be described in detail here. The center pixel coordinates of high-magnitude stars are retrieved based on the matching results and the high-magnitude star catalog. The range of star magnitudes selected for the high-magnitude star catalog should be smaller than the threshold. Obviously, the longer the star orbit, the smaller the magnitude threshold. Star orbit length Linear functions: (5) The coefficients of the above linear function , The coefficients in the above formula can be obtained by fitting test data from optical cameras, taking star charts of stars with different orbit lengths, analyzing the maximum magnitude that needs to be masked, and then fitting the data.

[0029] Step 4: Generate a star trail mask based on the star trail length, twist width, and star trail width.

[0030] The star trail mask is a rectangular mask, therefore the coordinates of the four corner points of the star trail mask need to be calculated, as follows: ; ; ; (6).

[0031] The coordinates of the four corner points are as follows: Figure 1 As shown.

[0032] Step 5: Perform star orbit rotation transformation using the star orbit slope.

[0033] The star trail mask obtained in step four is parallel to the image's X-axis and located at the image origin. It needs to be rotated and translated to the center pixel coordinates of the star. Rotation angle... Pick: (7) Rotation matrix for: (8) The final coordinates of the four corner points are: ; ; ; (9).

[0034] Step six: Repeat steps four and five to generate rectangular masks for all star points, and then merge the masks.

[0035] Repeat steps four and five to iterate through all star point coordinates. Use the four corner point coordinates generated by equation (9) to generate all star trail matrix masks and merge the masks as the final image mask to remove star trails.

[0036] Example 2 This embodiment further defines Embodiment 1. Taking the space target observation image captured by the Jilin-1 platform 02A01 satellite as an example, the implementation is carried out as described above: Following step one, a frame of image data is selected to calibrate the relationship between star trail width, star trail length, and star magnitude. Table 1 below shows the record of star magnitude and star trail width, using exponential fitting as follows: Figure 2 As shown.

[0037] Table 1: Record of Star Magnitude and Star Trail Width

[0038] The fitting formula is obtained as follows: (10) The fitting residual is 3.8, therefore We can take 11.4, and after appropriate rounding and expansion, we get 12. The star trail length of this frame is measured to be 70. This completes the calibration of the star trail width. The new image data star trail width is: (11) Based on step two, for the new image data, the image recognition algorithm can obtain the pixel coordinates of the four corner points and the center pixel coordinates of the star trails of stars with low magnitude (i.e., brighter stars), as shown in Table 2 (a total of 50 star trails were identified): Table 2. Identified star orbit coordinates

[0039] The length of the star trails was calculated by statistically truncating the mean. The twist width is 74.8. The slope is 9.8. It is 9.33.

[0040] According to step three, star map matching is performed using the center point coordinates, and the magnitude threshold for masking is calculated to be 14.42 according to equation (5). The number of stars with magnitude 14.42 in the field of view is 1057 using the high magnitude star catalog.

[0041] According to step four, calculate the coordinates of the four corner points of the star trail mask.

[0042] According to step five, the star trail mask is rotated and transformed.

[0043] Following step six, repeat steps four and five to mask each of the 1057 star trails individually, and then merge the masks. Use the masks to process the image, such as... Figure 3 and Figure 4 These are schematic diagrams of the images before and after processing, respectively.

Claims

1. A method for calculating star trail masks in space target observation image processing, characterized in that, The method includes: S1. The relationship between star orbit width, star orbit length, and star magnitude is determined by exponential fitting. S2. Provide the star trail length, distortion width, and slope of the image to be processed through sample statistics; S3. Use the identified low-magnitude stars to perform star map matching and invert the coordinates of high-magnitude stars; S4. Generate a star trail mask based on the star trail length, twist width, and star trail width; S5. Use the slope of star orbits to perform star orbit rotation transformations; S6, repeating steps S4 and S5, generates a rectangular mask for all star points, and merges the masks to create the final image mask for removing star trails.

2. The method for calculating star trail masks for space target observation image processing according to claim 1, characterized in that, The relationship between the width of a star orbit, its length, and the magnitude of the star is as follows: ;in, and Represents the fitted parameters, Indicates the width of the star trails. Indicates the length of the star orbit. This indicates the length of star trails in a standard star trail image. Indicates the magnitude of a star. This indicates that the star trail width is redundant.

3. The method for calculating star trail masks for space target observation image processing according to claim 2, characterized in that, Step S2 is as follows: For the image to be processed, the pixel coordinates of the four corner points and the center pixel coordinates of the star trails of low magnitude stars are obtained through image recognition algorithms. The length of each identified star trail is then calculated using the coordinates of the four corner points. and width ; Based on the identified star trail length The 10% truncation mean is used as the final star trail length. ; For the identified star trail width The mean of the 10% cutoff is used as the final twist width. ; For the star orbit slope, the mean value with a 10% cutoff is calculated as the final star orbit slope. .

4. The method for calculating star trail masks for space target observation image processing according to claim 3, characterized in that, The low-magnitude stars are those with relatively high brightness that can be extracted using image recognition methods and used for star map matching. The high-magnitude stars are those with relatively low brightness that cannot be extracted or are extracted inaccurately using image recognition methods. The classification of low-magnitude and high-magnitude stars is based on specific circumstances.

5. The method for calculating star trail masks for space target observation image processing according to claim 4, characterized in that, Step S3 specifically involves: using the star orbit center pixel coordinates obtained in step S2 to perform star map matching, and then retrieving the center pixel coordinates of high-magnitude stars based on the matching results and the high-magnitude star catalog. The range of star magnitudes selected for the high-magnitude star catalog is less than the threshold. threshold The length of the star orbit Linear functions: The coefficients in the above linear function , By fitting the test data of the optical camera and taking star charts of stars with different orbit lengths, the maximum magnitude that needs to be masked is analyzed, and then the coefficients in the above formula are fitted.

6. The method for calculating star trail masks for space target observation image processing according to claim 5, characterized in that, In step S4, the coordinates of the four corner points of the star trail mask are calculated. They are respectively: ; ； ； 。 7. The method for calculating star trail masks for space target observation image processing according to claim 6, characterized in that, Step S5 specifically involves selecting the rotation angle. , Rotation matrix for: Then the final coordinates of the four corner points for: ； ； ； 。 8. A star trail mask calculation system for space target observation image processing, characterized in that, The system includes: The unit that calibrates the relationship between star orbit width, star orbit length, and star magnitude is determined by exponential fitting; The unit values ​​of star trail length, distortion width, and slope of the image to be processed are given through sample statistics. A unit that uses identified low-magnitude stars to perform star map matching and invert the coordinates of high-magnitude stars; The star trail mask is generated based on the star trail length, twist width, and star trail width. A unit that uses the slope of star orbits to perform star orbit rotation transformations; The process iterates to generate rectangular masks for all star points, and then merges these masks to create the final image mask for removing star trails.

9. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1-7.

10. A computer-readable storage medium for storing computer instructions, characterized in that, When the computer instructions are executed by the processor, they implement the steps of the method according to any one of claims 1-7.

Images