High-precision orbit determination method for space targets observed by visible light

Abstract

This invention provides a high-precision orbit determination method for space targets using visible light observation, relating to the field of space target orbit determination technology. The method includes: S1, visible light image acquisition: acquiring raw visible light images of the space target and simultaneously recording the acquisition data information of the raw visible light images. By using image pixel restoration, the method allows the acquisition of raw images using low-cost, low-precision optical systems. Algorithms compensate for optical defects, significantly reducing payload cost and weight while ensuring orbit determination accuracy. This breaks through the dependence of traditional methods on high-precision optical hardware. Furthermore, combined with pre-calibrated bad pixel table compensation and fixed-mode noise removal, real-time processing is achieved, meeting the timeliness requirements of on-orbit orbit determination. Through the use of observation segment determination, observation segments are constructed by combining continuous image coordinates. Combined with high-precision image plane trajectory fitting, target trajectory extraction can be completed within a relatively short observation arc segment.

Description

Technical Field

[0001] This invention relates to the field of space target orbit determination technology, specifically a high-precision orbit determination method for space targets observed by visible light. Background Technology

[0002] Visible light observation, as an important means of monitoring space targets, has advantages such as long detection range, high measurement accuracy, and relatively low equipment cost. It has wide applications in fields such as space debris monitoring, non-cooperative target tracking, and spacecraft on-orbit servicing. Traditional visible light orbit determination methods usually rely on high-precision optical systems to acquire images, extract the target's centroid coordinates, and combine them with angle measurement information to determine the orbit.

[0003] However, existing technologies still have the following shortcomings:

[0004] 1. To ensure orbit determination accuracy, traditional methods typically employ high-precision, large-aperture optical lenses, which are difficult to manufacture, costly, and heavy, making it difficult to meet the payload requirements of low-cost platforms such as micro and nano satellites. At the same time, traditional image processing and orbit determination algorithms involve large computational loads, making it difficult to achieve real-time processing under limited resources.

[0005] 2. Traditional optical orbit determination methods rely on accumulating angle information over a long observation arc, which cannot achieve high-precision orbit determination within a short arc. For scenarios such as space debris monitoring, it is often necessary to quickly acquire target trajectory information within a limited observation window, and traditional methods are difficult to meet the timeliness requirements.

[0006] 3. Visible light observations are easily affected by factors such as cloud cover, atmospheric turbulence, sensor noise, and stray light interference, resulting in outliers in the observation data. Traditional trajectory fitting methods are sensitive to outliers, and the fitting results are easily affected by single-point errors, resulting in large deviations. It is difficult to obtain smooth and continuous trajectory routes under complex observation conditions, which in turn affects the accuracy and reliability of orbit determination. Summary of the Invention

[0007] The purpose of this invention is to provide a high-precision orbit determination method for space targets observed in visible light. By utilizing image pixel recovery, it allows the acquisition of raw images using low-cost, low-precision optical systems. Algorithms compensate for optical defects, significantly reducing payload cost and weight while maintaining orbit determination accuracy. This breaks through the dependence of traditional methods on high-precision optical hardware. Furthermore, combined with pre-processing operations such as pre-calibrated bad pixel table compensation and fixed-pattern noise removal, real-time processing is achieved, meeting the timeliness requirements of on-orbit orbit determination. By using observation segment determination, observation segments are constructed by combining continuous image coordinates. Combined with high-precision image plane trajectory fitting, target trajectory extraction can be completed within a short observation arc, providing technical support for scenarios such as space debris monitoring and rapid response to non-cooperative targets. By using orbit generation, the image plane trajectory is fitted by a B-spline curve model, and robust processing is performed using an M-estimation method based on the Huber function, effectively suppressing the influence of outliers in the observation data. Even with interference from cloud cover, sensor noise, etc., a smooth and continuous trajectory route can still be obtained.

[0008] To achieve the above effects, the present invention provides the following technical solution: a high-precision orbit determination method for space targets observed by visible light, comprising:

[0009] Step 1: Visible light image acquisition: Acquire raw visible light images of the space target and simultaneously record the acquisition data information of the raw visible light images.

[0010] It also includes the following steps:

[0011] Step 2, Image Pixel Restoration: Perform pixel compensation on the original visible light image to obtain a visible light enhanced image.

[0012] Step 3: Coordinate Acquisition: Analyze the enhanced visible light image to obtain the coordinates of the enhanced visible light image.

[0013] Step 4: Determine the observation segment: Combine the continuous visible enhanced light image coordinates to obtain the target image observation segment.

[0014] Step 5, Track Generation: Fit the coordinates of the continuous visible light enhanced image within the target image observation segment, and process outliers during the fitting process to obtain a smooth and continuous trajectory.

[0015] Furthermore, in step one, the collected data information includes the observation time, target attitude information, and camera built-in parameters.

[0016] Furthermore, in step two, the pixel compensation step includes:

[0017] S201. The original visible light image is processed to remove sensor fixed mode noise, flat field correction, dark current, and optical system vignetting effects.

[0018] S202. Identify areas of abnormal pixel loss caused by detector defects, transmission errors, or saturation;

[0019] S203. Based on a pre-calibrated bad pixel table, compensation is performed to restore missing or abnormal pixels;

[0020] S204. Improve the distinction between the target and the background to generate a visible light enhanced image.

[0021] Furthermore, in step two, the pixel compensation adopts the U-Net architecture, and the U-Net architecture training dataset consists of simulated degradation images and on-orbit measured images.

[0022] Furthermore, in step three, a motion continuity criterion is constructed based on the target displacement between adjacent frames, and a brightness consistency criterion is constructed based on the target brightness stability. Candidate targets that do not meet the motion continuity criterion and the brightness consistency criterion are eliminated.

[0023] Furthermore, in step five, the fitting of the visible light enhanced image coordinates adopts a B-spline curve model, which is expressed as:

[0024]

[0025] in, Let t be the position vector corresponding to parameter t. For k-th order B-spline basis functions, The parameters to be solved for the control points.

[0026] Furthermore, in step five, the outlier handling employs the M-estimation method based on the Huber function for robustness processing.

[0027] A computer program product that, when read by a computer, executes the method described in any of the preceding claims.

[0028] A computer-readable storage medium storing a processor-executable program, which, when executed by a processor, is used to perform the method as described in any of the preceding claims.

[0029] A high-precision orbit determination device for visible light observation of space targets includes a processor and a memory, wherein the memory includes a computer-readable storage medium storing a processor-executable program, which, when executed by the processor, is used to perform the method as described above.

[0030] This invention provides a high-precision orbit determination method for space targets observed by visible light, which has the following beneficial effects:

[0031] 1. By using image pixel recovery, it is possible to acquire original images using low-cost, low-precision optical systems. By using algorithms to compensate for optical defects, the cost and weight of the payload can be significantly reduced while ensuring orbit determination accuracy. This breaks through the dependence of traditional methods on high-precision optical hardware. Furthermore, with the pre-calibrated bad pixel table compensation and fixed-mode noise removal, real-time processing can be achieved to meet the timeliness requirements of on-orbit orbit determination.

[0032] 2. By using observation segments to determine the use of continuous image coordinates to construct observation segments, and combining them with high-precision image plane trajectory fitting, target trajectory extraction can be completed within a short observation arc, providing technical support for scenarios such as space debris monitoring and rapid response to non-cooperative targets.

[0033] 3. By using the trajectory generation method, the image plane trajectory is fitted by the B-spline curve model, and robust processing is performed by the M estimation method based on the Huber function. This can effectively suppress the influence of outliers in the observation data. Even with interference such as cloud cover and sensor noise, a smooth and continuous trajectory route can still be obtained. Attached Figure Description

[0034] Figure 1 This is a flowchart of a high-precision orbit determination method for space targets observed by visible light, according to the present invention.

[0035] Figure 2 This is a pixel compensation flowchart of a high-precision orbit determination method for space targets observed by visible light according to the present invention. Detailed Implementation

[0036] Please see Figure 1-2 This invention provides a technical solution: a high-precision orbit determination method for space targets observed by visible light, comprising:

[0037] Step 1: Visible light image acquisition: Acquire raw visible light images of the space target and simultaneously record the acquisition data of the raw visible light images.

[0038] It also includes the following steps:

[0039] Step 2, Image Pixel Restoration: Perform pixel compensation on the original visible light image to obtain an enhanced visible light image.

[0040] Step 3: Coordinate Acquisition: Analyze the visible light enhanced image to obtain the visible light enhanced image coordinates.

[0041] Step 4: Determine the observation segment: Combine the coordinates of continuous visible enhanced light images to obtain the target image observation segment.

[0042] Step 5, Track Generation: Fit the coordinates of continuous visible light enhanced images within the target image observation segment, and process outliers during the fitting process to obtain a smooth and continuous trajectory.

[0043] Specifically, in step one, the data collected includes the observation time, target attitude information, and camera built-in parameters.

[0044] Specifically, in step two, the pixel compensation steps include:

[0045] S201, Remove sensor fixed-mode noise, dark current, and optical system vignetting effects;

[0046] S202. Compensation is performed based on a pre-calibrated bad pixel table;

[0047] S203. Generate a visible light enhanced image.

[0048] Specifically, in step two, pixel compensation adopts the U-Net architecture, and the U-Net architecture training dataset consists of simulated degradation images and on-orbit measured images.

[0049] Specifically, in step three, a motion continuity criterion is constructed based on the target displacement between adjacent frames, and a brightness consistency criterion is constructed based on the target brightness stability. Candidate targets that do not meet the motion continuity criterion and the brightness consistency criterion are eliminated.

[0050] Specifically, in step five, the visible light enhanced image coordinates are fitted using a B-spline curve model, which is expressed as:

[0051]

[0052] in, Let t be the position vector corresponding to parameter t. For k-th order B-spline basis functions, The parameters to be solved for the control points.

[0053] Specifically, in step five, outlier handling employs the M-estimation method based on the Huber function for robustness processing.

[0054] At the same time, the present invention also proposes a computer program product that executes the method described above when the computer program product is read by a computer.

[0055] At the same time, the present invention also proposes a computer-readable storage medium storing a processor-executable program, characterized in that the processor-executable program, when executed by a processor, is used to perform the method as described above.

[0056] At the same time, the present invention also proposes a high-precision orbit determination device for visible light observation of space targets, which includes a processor and a memory, wherein the memory includes a computer-readable storage medium storing a processor-executable program, which is used to perform the method described above when executed by the processor.

[0057] This invention provides a high-precision orbit determination method for space targets observed in visible light, comprising the following steps: Step 1, Visible Light Image Acquisition: Acquire raw visible light images of the space target and simultaneously record the acquisition data information of the raw visible light images. The acquisition data information includes observation time, target attitude information, and camera built-in parameters. For faint targets, multi-frame non-destructive readout or long exposure can be used, but the trailing effect caused by target motion must be considered. If multiple devices are used simultaneously, a unified time reference and coordinate reference are required; Step 2, Image Pixel Restoration: Perform pixel compensation on the raw visible light image to obtain a visible light enhanced image. The pixel compensation steps include: S201, removing sensor fixed mode noise, flat field correction, dark current, and optical system vignetting effects from the raw visible light image; S202, identifying pixel abnormality missing areas caused by detector defects, transmission errors, and saturation; S203, compensating based on a pre-calibrated bad pixel table to restore missing and abnormal pixels; S204, improving the distinction between the target and the background to generate a visible light enhanced image. Points include dead pixels, hot pixels, and overexposed pixels. In a laboratory environment, multiple sets of dark-field and flat-field images with different exposures and temperatures are acquired. A bad pixel table is generated by statistically analyzing outliers. Bad pixel compensation employs neighborhood interpolation and multi-frame overlay or adjacent frame information restoration. Pixel compensation uses the U-Net architecture, with the U-Net architecture training dataset consisting of simulated degradation images and on-orbit measured images. The U-Net architecture is a classic encoder-decoder convolutional neural network architecture. Using the U-Net architecture for pixel compensation of space observation targets leverages the nonlinear mapping capabilities of deep learning to drive the inverse process of the physical degradation model. Skip connections preserve geometric accuracy, making it suitable for remote sensing mapping. Multi-scale features adapt to the coexistence of large and small celestial bodies. The skip connections in U-Net are the core, reducing positioning errors by approximately 30%–50% compared to purely fully connected networks. For space targets, both large-field targets and point targets may coexist; U-Net's multi-scale features can handle both types simultaneously. For on-orbit applications, a quantized and pruned lightweight version can be used, achieving inference speeds of 10⁻⁵ ms. FPS, Step 3, Coordinate Acquisition: Analyze the visible light enhanced image to obtain the visible light enhanced image coordinates. Construct a motion continuity criterion based on the target displacement between adjacent frames and a brightness consistency criterion based on the target brightness stability. Eliminate candidate targets that do not meet the motion continuity criterion and brightness consistency criterion. Adjust the ROI size according to the target brightness. Perform bicubic interpolation on the ROI and then calculate the centroid, which can reach 0.0.1 pixel-level accuracy: For extremely dark targets, a two-dimensional Gaussian surface fitting method is used, which is more accurate than the centroid method. Candidate targets that meet both motion continuity and brightness consistency are confirmed as the same target. Targets that do not meet either condition are marked as "false targets" and removed. Step 4: Determining the observation segment: Combine the coordinates of continuous visible light enhanced images to obtain the target image observation segment. The length of the target image observation segment should be no less than 5 to 10 frames of continuous tracking to ensure the stability of subsequent fitting. If the target is temporarily lost, a maximum number of allowed interruption frames can be set. If the target reappears near the predicted position after the interruption, it is considered as the same observation segment. Step 5: Trajectory generation: Fit the coordinates of continuous visible light enhanced images within the target image observation segment and process outliers during the fitting process to obtain a smooth and continuous trajectory. The fitting of the visible light enhanced image coordinates adopts a B-spline curve model, which is expressed as: ,in, Let t be the position vector corresponding to parameter t. For k-th order B-spline basis functions, For the parameters to be solved for the control points, outlier handling employs the M-estimation method based on the Huber function for robust processing. The B-spline curve model essentially uses parametric curve fitting / interpolation methods to solve problems in orbit representation, prediction, and data processing in astrodynamics. The observation data may contain outliers that have not been completely removed. M-estimation is used to reduce the weight of outliers. The Huber algorithm is a loss function that combines mean squared error and mean absolute error, aiming to balance the advantages and disadvantages of both. When the error is small, the Huber loss is similar to MSE, with smooth gradients, which helps to achieve fast convergence. When the error is large, the Huber loss switches to MAE, which is insensitive to outliers and reduces their interference with the model. By adjusting the hyperparameters, the emphasis of the loss function on MSE and MAE can be controlled.

[0058] In all the above embodiments, in order to achieve certain special data transmission and read / write function requirements, the above methods and corresponding devices can be expanded by adding devices, modules, components, hardware, pin connections or memory, processor differences during operation.

[0059] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the methods, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0060] In the several embodiments provided by the present invention, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of method steps is merely a logical or functional division. In actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the mutual coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, devices or units, and may be electrical, mechanical or other forms.

[0061] The units described as separate components of the various steps and devices in the method may or may not be logically or physically separated, and may not be physical units. That is, they may be located in one place or distributed across multiple network units. Some or all of the units may be selected to achieve the purpose of the solution in this embodiment according to actual needs.

[0062] In addition, the method steps and their implementations and functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in the form of hardware plus software functional units.

[0063] The above-described methods and apparatus can be implemented as integrated units in the form of software functional units, which can be stored in a computer-readable storage medium. The software functional units stored in the storage medium include several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute some steps of the methods of the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), NVRAM, magnetic disks, or optical disks.

[0064] The above description is only 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 within the protection scope of the present invention.

Claims

1. A high-precision orbit determination method for space targets observed by visible light, comprising: S1. Visible light image acquisition: Acquire raw visible light images of space targets and simultaneously record the acquisition data information of the raw visible light images; Its characteristic is that it further includes the following steps: S2. Image pixel restoration: Perform pixel compensation on the original visible light image to obtain a visible light enhanced image; S3. Coordinate Acquisition: Analyze the visible light enhanced image to obtain the visible light enhanced image coordinates; S4. Observation segment determination: Combine the continuous visible enhanced light image coordinates to obtain the target image observation segment; S5. Trajectory Generation: Fit the coordinates of the continuous visible light enhanced image within the target image observation segment, and process outliers during the fitting process to obtain a smooth and continuous trajectory.

2. The high-precision orbit determination method for space targets observed by visible light according to claim 1, characterized in that, In step S1, the acquired data information includes the observation time, target attitude information, and camera built-in parameters.

3. The high-precision orbit determination method for space targets observed by visible light according to claim 1, characterized in that, In step S2, the pixel compensation step includes: S201. The original visible light image is processed to remove sensor fixed mode noise, flat field correction, dark current, and optical system vignetting effects. S202. Identify areas of abnormal pixel loss caused by detector defects, transmission errors, or saturation; S203. Based on a pre-calibrated bad pixel table, compensation is performed to restore missing or abnormal pixels; S204. Improve the distinction between the target and the background to generate a visible light enhanced image.

4. The high-precision orbit determination method for space targets observed by visible light according to claim 1, characterized in that, In step S2, the pixel compensation adopts the U-Net architecture, and the U-Net architecture training dataset consists of simulated degradation images and on-orbit measured images.

5. The high-precision orbit determination method for space targets observed by visible light according to claim 1, characterized in that, In step S3, a motion continuity criterion is constructed based on the target displacement between adjacent frames, and a brightness consistency criterion is constructed based on the target brightness stability. Candidate targets that do not meet the motion continuity criterion and the brightness consistency criterion are eliminated.

6. The high-precision orbit determination method for space targets observed by visible light according to claim 1, characterized in that, In step S5, the fitting of the visible light enhanced image coordinates adopts a B-spline curve model, which is expressed as: in, Let t be the position vector corresponding to parameter t. For k-th order B-spline basis functions, The parameters to be solved for the control points.

7. The high-precision orbit determination method for space targets observed by visible light according to claim 1, characterized in that, In step S5, the outlier handling employs the M-estimation method based on the Huber function for robustness processing.

8. A computer program product, characterized in that, When the computer program product is read by a computer, the method described in any one of claims 1 to 7 is executed.

9. A computer-readable storage medium storing a processor-executable program, characterized in that, The processor-executable program, when executed by the processor, is used to perform the method as described in any one of claims 1-7.

10. A high-precision orbit determination device for space targets observed in visible light, characterized in that, It includes a processor and a memory, wherein the memory includes a computer-readable storage medium storing a processor-executable program, which, when executed by the processor, is used to perform the method as described in any one of claims 1-7.

Images