Three-axis stable space target attitude estimation method

Abstract

The invention discloses a three-axis stable space target attitude estimation method. The method comprises the following steps: extracting a target contour of each frame of image from a sequence ISAR (Inverse Synthetic Aperture Radar) image; matching the extracted target contour with a pre-generated template library contour, and outputting an optimal matching attitude list of each frame; calculating an imaging projection matrix of each frame according to the number of target orbits and the position of a ground observation station, and performing attitude conversion on the attitude list of each frame to obtain the converted attitude list of each frame of the satellite-based orbital coordinate system; performing attitude ambiguity resolution in combination with each frame of attitude sequence, and outputting a unique attitude as a coarse attitude estimation result; and taking a coarse attitude estimation result as an initial value, carrying out optimization search on a neighborhood of the result, taking a three-dimensional attitude of a target model as an optimization variable, optimizing an average Hausdorff distance of a theoretical projection contour and an actual imaging contour to be minimum, and outputting an accurate target attitude. According to the invention, high-precision and high-efficiency attitude estimation of the three-axis stable space target can be realized.

Description

Technical Field

[0001] The present invention belongs to the field of target recognition based on ISAR images, and more particularly, relates to a method for estimating the attitude of a three-axis stabilized space target. Background Art

[0002] With the rapid development of space technology, effective space target state estimation and collision avoidance strategies are particularly important for ensuring space safety. The on-orbit attitude of a space target, as a key parameter, is crucial for determining the target's motion state. By estimating the attitude of a space target, the working conditions of cooperative targets can be diagnosed and analyzed, and potential abnormal behaviors of non-cooperative targets can also be identified. In addition, predicting the attitude of a target helps to predict its azimuth change and make corrections, prevent possible collisions with space debris, and accurate attitude estimation is also crucial for performing maintenance tasks and spacecraft docking. Compared with optical sensors, ISAR (Inverse Synthetic Aperture Radar) has the ability to operate all day and all weather, avoiding the limitations of day and night and weather, so it has significant advantages in the field of space target attitude estimation.

[0003] Most of the existing ISAR-based methods for estimating the attitude of three-axis stabilized space targets start from the target model, extract features such as the satellite body and solar panels, and then perform feature recognition. These methods have the advantages of intuitive geometric features and clear attitude interpretation methods, and are the main methods for space target attitude estimation. However, the difficulty of extracting target component features leads to the need to improve the efficiency and accuracy of attitude estimation algorithms. Summary of the Invention

[0004] The purpose of the present invention is to provide a method for estimating the attitude of a three-axis stabilized space target, a computer device, a computer-readable storage medium, and a computer program product, which can solve the problem of low estimation efficiency and accuracy of existing methods.

[0005] To achieve the above object, one aspect of the present invention provides a method for estimating the attitude of a three-axis stabilized space target, including: Step S1: Based on the Clean algorithm, extract the target contour of each frame of image from the sequence of ISAR images; Step S2: Match the extracted target contours with the pre-generated template library contours respectively, use the Hausdorff distance as the matching operator, and output a list composed of the top several attitudes with the smallest distance for each frame; Step S3: According to the target orbit elements and the position of the ground observation station, calculate the imaging projection matrix for each frame, and perform attitude conversion on each frame of attitude list to obtain the converted attitude list for each frame in the satellite-based orbital coordinate system; Step S4. For each frame of the obtained converted pose list, extract a candidate pose from each frame, traverse all possible pose sequences, and find the pose sequence with the minimum variance as the rough pose estimation result; Step S5. Using the rough pose estimation result as the initial value, perform an optimized search in the neighborhood of this result. Taking the three-dimensional pose of the target model as the optimization variable, optimize the average Hausdorff distance between its theoretical projection contour and the actual imaging contour to the minimum, and output the accurate target pose.

[0006] Preferably, the said Step S1 includes: Step S11. Extract all maximum scattering points of each frame of image based on the Clean algorithm; Step S12. Use the connectivity analysis method to remove the noise points of the point set obtained in Step S11; Step S13. For the point set after removing the noise points in Step S12, use Alpha-Shape and wavelet filtering to obtain the target contour.

[0007] Preferably, the template library in Step S2 is generated as follows: Assume that the positions of each point of the target at zero pose are , then the position of the target point cloud at any pose is:

[0008] where, , , respectively represent the yaw, pitch, and roll angles, is the rotation matrix for rotating by an angle around the axis, is the rotation matrix for rotating by an angle around the axis, is the rotation matrix for rotating by an angle around the axis; Traverse , , , establish the target positions at all poses, project the three-dimensional model of the target at each pose onto the two-dimensional plane to form a two-dimensional closed contour, thereby generating the template library.

[0009] Preferably, the said Step S3 includes: Step S31. Construct the projection matrix:

[0010] where, and are the projection vectors for the distance and azimuth dimensions respectively, and are unit vectors in the range and azimuth dimensions respectively, and are the range and azimuth resolutions respectively; Step S32: According to the target orbital elements and the position of the ground observation station, calculate the radar line of sight at the middle moment of single-frame imaging as the unit vector in the range dimension, obtain the range resolution through the radar parameters, calculate the azimuth unit vector and azimuth resolution by calculating the single-frame LOS rotation angle, so as to obtain the projection matrix ; Step S33: Perform image plane registration on the attitude list of each frame to achieve the conversion from the index attitude of the template to the actual attitude

[0011] Preferably, in the step S33, use the following attitude conversion formula to achieve the conversion from the index attitude of the template to the actual attitude

[0012] where, is the quaternion required to rotate from the zero attitude target to the actual attitude vector expansion of , where, respectively represent the azimuth axis rotation quaternion, the range axis rotation quaternion, and the quaternion obtained by converting from the rotation matrix corresponding to the index attitude ;

[0013] Preferably, the step S4 includes: Step S41: For the converted attitude list obtained in step S3, take the top several attitudes with the best match for each frame; Step S42: Adopt the attitude angle minimum Euclidean distance method, select an attitude from the first frame, find the attitude closest to it in the second frame, and use this method until the last frame to form an attitude sequence, and calculate the sequence variance; Step S43: Traverse the several attitudes selected in the first frame, and repeat step S42 to obtain several sequences and variances; Step S44: In the sequences in step S43, take the sequence with the minimum variance, and calculate the mean value to obtain the rough attitude estimation of the target.

[0014] Preferably, in the step S5, adopt a PID-based search algorithm, and the optimization function is:

[0015] Among them, is the actual ISAR image of the th frame, is the projection matrix of the th frame, is the position of the zero-attitude target, is the attitude rotation matrix, is the contour extraction function, is the function for calculating the Hausdorff distance between two contours, , , , among which, is the rough attitude estimation result.

[0016] Another aspect of the present invention provides a computer device, including a memory, a processor, and a computer program stored on the memory, and the processor executes the computer program to implement the steps of the above method.

[0017] Another aspect of the present invention provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the above method are implemented.

[0018] Another aspect of the present invention provides a computer program product, including a computer program, and when the computer program is executed by a processor, the steps of the above method are implemented.

[0019] According to the three-axis stabilized space target attitude estimation method, computer device, computer-readable storage medium, and computer program product of the above aspects of the present invention, high-precision and high-efficiency attitude estimation of the three-axis stabilized space target can be achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions of the present invention, the drawings used in the description of the embodiments of the present invention will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention, and those of ordinary skill in the art can also obtain other drawings based on these drawings without creative efforts: Figure 1 is a flowchart of the three-axis stabilized space target attitude estimation method according to an embodiment of the present invention; Figure 2 is a schematic diagram of the three-axis stabilized space target attitude estimation method according to an embodiment of the present invention; Figure 3 is an example diagram of attitude defuzzification by combining attitude sequences of each frame according to an embodiment of the present invention; Figure 4 is a structural diagram of a computer device according to an embodiment of the present invention. Detailed implementation manners

[0021] To make the objectives, technical solutions and advantages of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. Apparently, the described embodiments are only a part rather than all of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0022] An embodiment of the present invention provides a method for estimating the attitude of a three-axis stabilized space target, as Figure 1 shown. The method of the embodiment of the present invention includes steps S1 to S5. Taking a certain radar simulation data as an example below, combined with Figure 2 the schematic diagram, each step of the method of the embodiment of the present invention will be described in detail. In this embodiment, the radar system simulation parameters are as follows: the radar frequency bandwidth is 700 MHz, the wavelength is 0.05 m, the pulse repetition frequency is 50 Hz, the pulse width is 4 , the radar emits a linear frequency modulation signal, generates five frames of images at equal intervals in a single circle observation, the slow-time echo accumulation number is 256, and the single echo signal-to-noise ratio is 10 dB.

[0023] Step S1: Extract the spatial target contour from the sequence of ISAR images.

[0024] In this step, an ISAR image target contour extraction method based on the Clean algorithm improved by the number of iterations is used to extract the target contour of each frame of image.

[0025] Further, step S1 includes steps S11 to S13.

[0026] Step S11: Record the maximum scattering points in each frame of image, subtract the point spread function (PSF) from the image, and record the positions of all the maximum scattering points in the image during the iterative function recording process.

[0027] In this embodiment, the number of iterations uses a surrogate optimization algorithm to select the optimal threshold, extract as many scattering points of the target body as possible, and do not introduce too much noise. The cost function is:

[0028] Wherein:[[]]END]] is the threshold to be optimized, that is, the number of extraction points of the Clean algorithm, is the point on the extracted target, is the point outside the extracted target, that is, noise, is the weight of the noise, which is set to 0.1 in this embodiment.

[0029] Step S12. Use the connectivity analysis method to remove the noise points in the point set obtained in Step S11.

[0030] In this embodiment, first, the isolated points in the image are removed. If the number of points in the neighborhood of each point is less than 3, it is determined as an isolated point. Then, the largest connected component is retained to remove the noise aggregation points. The point sets with an adjacent distance less than 2 are divided into one connected component, and the connected component with the largest number of points is retained.

[0031] Step S13. For the point set output in Step S12, use Alpha-Shape and wavelet filtering to obtain the target contour.

[0032] In this embodiment, the circle radius of the Alpha-Shape algorithm is set to 7, and the wavelet basis function is set to the db wavelet.

[0033] Step S2. Match the target contour of each frame of image with the template library and output the list of the optimal matching poses for each frame.

[0034] Specifically, match the target contour extracted in Step S1 with the pre-generated template library contours respectively, use the Hausdorff distance as the matching operator, and output a list composed of the top several poses with the minimum distance for each frame.

[0035] Furthermore, the method for generating the template library in Step S2 is as follows. Assume that the positions of each point of the target in the zero pose are , then the position of the target point cloud in any pose is:

[0036] where , , respectively represent the yaw, pitch, and roll angles, is the rotation matrix around the axis by the rotation angle , is the rotation matrix around the axis by the rotation angle , is the rotation matrix around the axis by the rotation angle . Traverse , , , establish the target positions in all poses, project the three-dimensional model of the target in each pose onto the two-dimensional plane to form a two-dimensional closed contour, which is the generated template library. In this embodiment, the reference projection plane is selected as the reference distance axis and the reference azimuth axis , and the distance and azimuth resolutions are both set to .

[0037] Step S3: Calculate the imaging projection matrix for each frame according to the target orbital elements and the positions of ground observation stations, and perform attitude conversion on each frame of the attitude list to obtain the converted attitude list in the satellite-based orbital coordinate system for each frame.

[0038] In this step, according to the target orbital elements and the positions of ground observation stations, the radar line of sight (LOS) at the intermediate moment of single-frame imaging is calculated as the range unit vector; the range resolution is obtained through radar parameters; the azimuth unit vector and azimuth resolution are obtained by calculating the LOS rotation angle of a single frame.

[0039] For the attitude list obtained in step S2, use quaternions to rotate the range and azimuth dimension vectors of the template imaging plane to completely align them with the two-dimensional vectors of the actual imaging plane. At this time, the target under the index attitude following the rotation of the template imaging plane will also finally align with the actual attitude, and the converted attitude list is output.

[0040] Furthermore, step S3 includes steps S31 to S33: Step S31: Construct the projection matrix:

[0041] where and are the range and azimuth dimension projection vectors, and are the range and azimuth dimension unit vectors, and are the range and azimuth resolutions, which can be calculated from radar parameters, is the signal bandwidth of the radar, is the speed of light.

[0042] Step S32: Through coordinate transformation, obtain the LOS sequence directions in the target orbital coordinate system for each frame. The range dimension direction can be regarded as the LOS direction at the imaging center moment; the azimuth dimension direction is perpendicular to the LOS rotation axis and the range dimension direction; the azimuth resolution , where the imaging accumulation angle can be obtained from the LOS rotation angle during the accumulation time. Thus, the projection matrix can be obtained.

[0043] Step S33: Perform image plane registration on the attitude list obtained in step S2 for each frame to achieve the conversion from the index attitude of the template to the actual attitude .

[0044] Furthermore, the specific calculation method of step S33 is as follows: Construct the quaternion Distance dimension of the template Obtained after quaternion rotation , which has been aligned with the actual distance dimension . This rotation is also applied to the azimuth dimension of the template and the target in the indexed pose . At this time, one axis is rotationally aligned; the azimuth axis needs to be aligned again, and a quaternion is constructed using the same method . After two rotations, the alignment of the two image planes can be achieved, and at the same time, the template indexed pose target is aligned with the actual pose target. Next, the actual pose value needs to be solved

[0045] Let the matching pose index be , and the corresponding rotation matrix be . It can be obtained that

[0046] The target in the indexed pose is

[0047] where is the zero-pose target; the rotation matrix is converted to a quaternion . According to the multiplicability of quaternion rotation, to rotate from the zero-pose target to the actual pose, the quaternion

[0048] is required, where respectively represent the quaternion for azimuth axis rotation, the quaternion for distance axis rotation, and the quaternion converted from the rotation matrix determined by the matching pose. is the vector expansion of the quaternion. Applying the quaternion to the zero-pose target gives the target after pose conversion, that is

[0049] From the relationship between quaternion and Euler angle, we have

[0050] The above formula is the pose conversion formula, which realizes the conversion from the indexed pose of the template to the actual pose , and solves the pose mismatch problem

[0051] Step S4: Combine the pose sequences of each frame to perform pose deblurring and output a unique pose

[0052] For the list of converted poses of each frame obtained in step S3, a candidate pose is taken from each frame, and all possible pose sequences are traversed to find the pose sequence with the smallest variance, which is the rough pose estimation result

[0053] An example of the step S4 in this embodiment is as follows Figure 3 shown. In this embodiment, the number of candidate postures selected for each frame is 5. It should be noted that at low signal-to-noise ratios, the number of candidate postures for each frame needs to be increased to ensure that the true posture appears in the candidates, sacrificing computational efficiency for accuracy.

[0054] In this embodiment, the step S4 includes steps S41 to S44: Step S41: For the converted posture list obtained in step S3, take the top several postures with the best match for each frame. For example Figure 3 shown, the top 5 candidate postures for the first frame are postures 0 to 4.

[0055] Step S42: Using the minimum Euclidean distance method of posture angles, select a posture from the first frame, find the posture in the second frame that is closest to it, and so on until the last frame to form a posture sequence, and calculate the sequence variance.

[0056] Step S43: Traverse the several postures taken from the first frame, and repeat step S42 to obtain several sequences and variances.

[0057] Step S44: Among the sequences in step S43, take the sequence with the minimum variance, and find the mean value to obtain the rough posture estimate of the target.

[0058] Step S5: Use an optimization algorithm to refine the estimate of the posture output in the previous step, and output the accurate target posture.

[0059] Specifically, using the rough posture estimate obtained in step S4 as the initial value, perform an optimization search on the neighborhood of this result. Directly use the three-dimensional posture of the target model as the optimization variable, and optimize the average Hausdorff distance between its theoretical contour and the actual five-frame imaging contour to the minimum, and output the refined posture result. That is, without relying on the template library, directly use the posture angle as the optimization variable to rotate the spatial target model, and then project to generate the theoretical contour, minimizing the difference between the theoretical projection contour and the actual contour. At this time, the posture angle is the optimal one.

[0060] Furthermore, the step S5 adopts a PID-based search algorithm (PSA). This algorithm shows good optimization ability in constrained optimization problems and has advantages in terms of computational cost and complexity; the optimization function is

[0061] wherein is the actual ISAR image of the th frame, is the projection matrix calculated for this frame; is the position of the zero-attitude target, is the attitude rotation matrix, is the contour extraction function, is the function for calculating the Hausdorff distance between two contours. The feasible region of this constrained optimization problem is , , , where is the rough attitude estimation result.

[0062] In this embodiment, the initial population number of PSA is set to 50, and the number of iterations is set to 100 times.

[0063] In this embodiment, for the target with the set attitude of , the method of this embodiment of the present invention outputs the result of , the Euler angle rotation error is , the calculation time is 113s; the error of the traditional shape context matching method is , the calculation time is 207s; the method of this embodiment of the present invention is superior to the traditional method in both accuracy and efficiency.

[0064] In summary, the three-axis stabilized space target attitude estimation method of this embodiment of the present invention is a method based on ISAR image target contour matching. It innovatively proposes an image plane registration algorithm, derives the image plane registration expression in detail, aligns different frame image planes with the template image plane to obtain the actual attitude of different frame template matches; then combines the actual attitudes of multiple frames to solve the multi-solution problem, and uses the PSA algorithm to efficiently obtain the precise attitude of the target. The method of this embodiment of the present invention can be used for three-axis stabilized targets with known models to achieve attitude estimation with high accuracy and low computer resource consumption, and has high engineering application value.

[0065] This embodiment of the present invention also provides a computer device, which can be a server, and its internal structure diagram can be as Figure 4 shown. This computer device includes a processor, a memory, and a network interface connected through a system bus. Among them, the processor of this computer device is used to provide computing and control capabilities. The memory of this computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of this computer device is used to store the operation parameter data of each frame. The network interface of this computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, it realizes the steps of the method of this embodiment of the present invention.

[0066] Those skilled in the art can understand,Figure 4 The structure shown is only a block diagram of some of the structures related to the solution of this application, and does not constitute a limitation on the computer device to which the solution of this application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0067] An embodiment of the present invention also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the method according to the embodiment of the present invention are implemented.

[0068] An embodiment of the present invention also provides a computer program product, including a computer program. When the computer program is executed by a processor, the steps of the method according to the embodiment of the present invention are implemented.

[0069] Only some exemplary embodiments of the present invention have been described above by way of illustration. Without doubt, for those of ordinary skill in the art, the described embodiments can be modified in various different ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and description are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. A three-axis stabilized space target attitude estimation method, characterized in that Including: Step S1: Based on the Clean algorithm, extract the target contours of each frame of the ISAR image sequence. Step S2: Match the extracted target contours with the pre-generated template library contours respectively, use the Hausdorff distance as the matching operator, and output a list composed of the top several poses with the smallest distance for each frame. Step S3: According to the target orbital elements and the positions of ground observation stations, calculate the imaging projection matrix for each frame, and perform pose transformation on each frame's pose list to obtain the transformed pose list of each frame in the satellite-based orbital coordinate system. Step S4: For the obtained transformed pose list of each frame, take a candidate pose from each frame, traverse all possible pose sequences, and find the pose sequence with the smallest variance as the rough pose estimation result. Step S5: Using the rough pose estimation result as the initial value, perform an optimization search in the neighborhood of this result, use the three-dimensional pose of the target model as the optimization variable, and optimize the average Hausdorff distance between its theoretical projection contour and the actual imaging contour to the minimum, and output the accurate target pose.

2. The method according to claim 1, characterized in that, The said Step S1 includes: Step S11: Based on the Clean algorithm, extract all maximum scattering points of each frame of the image. Step S12: Use the connectivity analysis method to remove the noise points in the point set obtained in Step S11. Step S13: For the point set after removing noise points in Step S12, use Alpha-Shape and wavelet filtering to obtain the target contour.

3. The method according to claim 1 or 2, characterized in that, Generate the template library in Step S2 as follows: Assume that the positions of each point of the target at zero pose are , then the position of the target point cloud at any pose is: Among them, , , respectively represent the yaw, pitch, and roll angles, is the rotation matrix for the rotation angle around the axis , is the rotation matrix for the rotation angle around the axis , is the rotation matrix for the rotation angle around the axis ; Traverse , , , establish the target positions in all poses. For the target 3D model in each pose, project it onto a 2D plane to form a 2D closed contour, thereby generating a template library.

4. The method according to claim 1 or 2, characterized in that, The said Step S3 includes: Step S31: Construct the projection matrix: wherein, and are respectively the projection vectors in the range and azimuth dimensions, and are respectively the unit vectors in the range and azimuth dimensions, and are respectively the range and azimuth resolutions; Step S32: Based on the target orbital elements and the positions of ground observation stations, calculate the radar line of sight at the intermediate moment of a single-frame imaging as the unit vector in the range dimension, obtain the range resolution through radar parameters, and calculate the azimuth unit vector and azimuth resolution through the single-frame LOS rotation angle, thereby obtaining the projection matrix ; Step S33: Perform image plane registration on the pose lists of each frame to achieve the conversion from the indexed pose of the template to the actual pose transformation.

5. The method according to claim 4, wherein In the step S33, the following attitude conversion formula is used to achieve the conversion from the index attitude of the template to the actual attitude as follows: wherein, is the quaternion required to rotate from the zero-attitude target to the actual attitude vector expansion of: ， Among them, respectively represent the azimuth axis rotation quaternion, the distance axis rotation quaternion, and the quaternion obtained by converting the rotation matrix corresponding to the index attitude .​ 6. The method according to claim 1 or 2, characterized in that, The said Step S4 includes: Step S41: For the transformed pose list obtained in Step S3, take the top several poses with the highest matching rank for each frame. Step S42: Adopt the minimum Euclidean distance method of pose angles, select a pose from the first frame, find the pose closest to it in the second frame, and use this method until the last frame to form a pose sequence and calculate the sequence variance. Step S43: Traverse the several poses taken from the first frame, and repeat Step S42 to obtain several sequences and variances. Step S44: Select the sequence with the smallest variance from the sequences in Step S43, and calculate the mean value to obtain the rough pose estimation of the target.

7. The method according to claim 1 or 2, characterized in that, In the said Step S5, adopt a search algorithm based on PID, and the optimization function is: Among them, is the actual ISAR image of the th frame, is the projection matrix of the th frame, is the position of the zero-attitude target, is the attitude rotation matrix, is the contour extraction function, is the function for calculating the Hausdorff distance between two contours, , , , among which, is the rough attitude estimation result.

8. A computer device, comprising a memory, a processor, and a computer program stored on the memory, characterized in that, The processor executes the computer program to implement the steps of the method described in any one of claims 1-7.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the method described in any one of claims 1-7.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the method described in any one of claims 1-7.

Images