A method, apparatus, and storage medium for multi-extended target measurement and segmentation based on OPTICS-FCM

Abstract

This invention proposes a multi-extended target measurement partitioning method, device, and storage medium based on OPTICS-FCM, relating to the field of multi-target tracking technology. It addresses the problems in existing multi-extended target tracking processes under dense clutter environments, such as ineffective clutter removal during measurement partitioning, triggering secondary measurement partitioning, measurement partitioning errors, and failure of neighboring target measurement partitioning, ultimately leading to incorrect target number estimation. The implementation steps include: preprocessing extended target measurement data to obtain parameters for OPTICS initialization parameter values; removing clutter measurements; outputting reachable distances and preliminary measurement subsets; estimating the measurement generation rate; and performing FCM measurement sub-partitioning based on the measurement rate to complete the final extended target measurement partitioning. This invention optimizes the OPTICS algorithm and FCM sub-partitioning based on extended target measurement data and can be used in the measurement partitioning stage of radar multi-extended target tracking.

Description

Technical Field

[0001] This invention relates to the field of multi-target tracking technology, and in particular to a multi-extended target measurement and segmentation method, apparatus and storage medium based on OPTICS-FCM. Background Technology

[0002] With the rapid development of radar technology and the continuous improvement of radar resolution, target echoes are no longer confined to a single resolution cell, but gradually expand to multiple resolution cells. These targets are called extended targets, and the standard measurement models assumed by conventional multi-target tracking methods are no longer suitable for tracking extended targets. The current mainstream method for tracking multiple extended targets is to first divide the measurements of multiple extended targets and then track them. Therefore, measurement division is a crucial step in the extended target tracking process.

[0003] Currently, classic multi-extended target measurement set partitioning algorithms include distance partitioning, K-means++ partitioning, prediction partitioning, and Expectation Maximization (EM) partitioning. These methods cannot effectively eliminate clutter, especially in dense ocean clutter environments. Strong sea clutter can interfere with measurement partitioning. In box particle filtering algorithms, the target measurement box is enlarged, severely impacting the accuracy of target number estimation in tracking methods. Furthermore, because the number of extended target measurements is uncertain, triggering fixed-threshold measurement sub-partitioning can easily lead to measurement partitioning errors and incorrect target number estimation.

[0004] Density-based partitioning methods are also common multi-extended target measurement set partitioning algorithms. For example, in the DBSCAN (Density-Based Spatial Clustering of Applications with Noise) clustering algorithm, the initial parameter values ​​ε-neighborhood and MinPts need to be manually set. Different parameter values ​​may lead to huge differences in measurement partitioning results. In this case, it cannot adapt to the real-time changing clutter environment, resulting in the inability to effectively remove clutter and affecting the tracking effect of the next extended target. Summary of the Invention

[0005] In view of this, the present invention proposes a multi-extended target measurement partitioning method, device and storage medium based on OPTICS (Ordering Points to identify the clustering structure)-FCM (Fuzzy C-Means) to solve the problems in existing multi-extended target tracking processes under dense clutter environments, such as measurement partitioning failing to effectively remove clutter, triggering secondary measurement partitioning, measurement partitioning errors, and failure of measurement partitioning of neighboring targets, ultimately leading to incorrect target number estimation.

[0006] Therefore, the present invention provides the following technical solution:

[0007] This invention discloses a multi-extended target measurement partitioning method based on OPTICS-FCM clustering, comprising the following steps:

[0008] Obtain a multi-extended target measurement set at a preset time, preprocess the multi-extended target measurement set, calculate the distance between each pair of measurement samples, and obtain the maximum distance.

[0009] Set the minimum number of samples in the neighborhood of the core point to be the measurement sample;

[0010] The measurement data, maximum distance, and minimum number of neighborhood samples in the multi-extended target measurement set are input into the OPTICS clustering algorithm. The maximum distance is used as the initial reachable distance for all measurement samples to obtain an ordered output result and the corresponding reachable distance.

[0011] Traverse the reachable distance sequence of the measurement samples, perform threshold judgment on the reachable distance data of the measurement samples, and complete the clutter removal;

[0012] The measurement rate estimation is completed by using the average number of samples in each subset as the measurement rate estimate.

[0013] Based on the measurement rate estimation, each measurement subset is subdivided. If the measurement subdivision conditions are met, FCM is used for measurement subdivision, and then the final measurement subdivision subset is output. The measurement subdivision conditions include: the measurement subset is composed of intersecting or adjacent targets.

[0014] Furthermore, the measurement data, maximum distance, and minimum number of neighborhood samples in the multi-extended target measurement set are input into the OPTICS clustering algorithm. The maximum distance is used as the initial reachability distance for all measurement samples, resulting in an ordered output and corresponding reachability distances, including:

[0015] Step 1: Calculate the core distance of the core point. In the neighborhood of the current core point o, the sample point with the smallest number of samples in the ascending order of the distance to the core point o is used as the core distance of o.

[0016] Step 2: Create two queues: a queue to be processed (QP) and a queue of results (order). The queue to be processed stores samples in the neighborhood of the core sample and their reachability distances, and sorts them in ascending order of reachability distance. The queue of results stores the output order of the sample points, which is the data that has been processed.

[0017] Step 3: If all points in the multi-expanded target measurement set have been processed or no core points exist, the algorithm terminates; otherwise, select an unprocessed sample point o that is a core point, first add o to the result queue order, and then remove o from QP; then find the measurement set Z. k All density values ​​of o in the data are directly connected to sample point x. Calculate the reachable distance from x to o. If x is not in the queue QP, put x and its reachable distance into QP. If x is in QP, update the reachable distance of x if the new reachable distance of x is smaller. Finally, reorder the data in QP according to the reachable distance from smallest to largest.

[0018] Step 4: If the queue QP is empty, return to step 3; otherwise, retrieve the first sample point y from QP, add it to the order, and add it to the measurement set Z. k The marking has been processed. The marking method is to mark the corresponding position of the sequence to be processed at position o as empty, and at the same time push o into the result queue order.

[0019] Step 5: If y is not a core point, repeat step 4, that is, find the sample point with the smallest reachable distance of the remaining data in QP; if y is a core point, find all density reachable sample points of y in the measurement set, calculate the reachable distance to y, and then update all density reachable sample points in QP according to step 3.

[0020] Step 6: Repeat steps 3 and 4 until all measurement samples have been processed and the waiting queue QP is emptied, then terminate the iteration; finally, an ordered output result and the corresponding reachable distance are obtained.

[0021] Furthermore, the reachability sequence of the measurement samples is traversed, and a threshold judgment is performed on the reachability data of the measurement samples, including:

[0022] Set the number of subsets, count, initially 0. Iterate through the reachable distance sequence RD of the measurement samples. Perform a threshold judgment on RD(order(i)). If RD(order(i)) is not less than RDTH and RD(order(i+1)) is less than RDTH, then the count value is increased by 1, and the sample output corresponding to the RD value before the next RD value that meets the condition is the measurement partition subset Zp. k (1), Zpk(1)=Z k (order(i)), after the RD traversal is complete, output each measurement subset Zp. k (p), and the number of subsets count, which is the estimated number of expanded targets. The measurement subset samples no longer contain clutter samples, thus completing clutter removal.

[0023] Furthermore, based on the aforementioned measurement rate estimation, each measurement subset is subdivided. Those meeting the subdivision criteria are subdivided using FCM, including:

[0024] Based on the estimated measurement rate, the number of measurements in each measurement subset is determined. If the number of measurement data columns in a measurement subset is greater than 1.5L... k Then, FCM measurement subdivision is performed on the measurement subset.

[0025] Furthermore, the distance is the Euclidean distance.

[0026] Furthermore, the minimum number of samples in the neighborhood is set to 5.

[0027] The present invention also provides a multi-extended target measurement partitioning device based on OPTICS-FCM clustering, comprising:

[0028] The preprocessing unit is used to acquire a set of multiple extended target measurements at a preset time, preprocess the set of multiple extended target measurements, calculate the distance between pairs of measurement samples, and obtain the maximum distance.

[0029] The parameter setting unit is used to set the minimum number of samples in the neighborhood where the measurement sample becomes the core point.

[0030] The OPTICS clustering unit is used to input the measurement data, maximum distance, and minimum number of neighborhood samples obtained by the parameter setting unit from the multi-extended target measurement set obtained by the preprocessing unit into the OPTICS clustering algorithm. The maximum distance is used as the initial reachable distance of all measurement samples to obtain an ordered output result and the corresponding reachable distance.

[0031] The clutter removal unit is used to traverse the reachability sequence of the measurement samples obtained by the OPTICS clustering unit, perform threshold judgment on the reachability data of the measurement samples, and complete clutter removal.

[0032] The measurement rate estimation unit is used to estimate the measurement rate by taking the average number of samples in each subset after the clutter removal unit removes clutter as the measurement rate estimate.

[0033] The FCM partitioning unit is used to perform sub-partitioning judgment on each measurement subset based on the measurement rate estimation obtained by the measurement rate estimation unit. If the measurement sub-partitioning conditions are met, FCM is used for measurement sub-partitioning, and then the final measurement partitioning subset is output.

[0034] The present invention also provides a computer-readable storage medium storing a computer instruction set, which, when executed by a processor, implements the multi-extended target measurement partitioning method based on OPTICS-FCM clustering as described above.

[0035] Advantages and positive effects of the present invention:

[0036] In this invention, measurement data is preprocessed to generate operable data processing conditions, which can adaptively adjust the OPTICS algorithm settings to improve the accuracy of measurement partitioning. Measurement data marked as clutter can be removed, preventing erroneous triggering of measurement sub-partitioning and avoiding tracking errors caused by excessively large measurement bin partitions in box particle filtering-based tracking methods. Furthermore, after clutter removal, the measurement generation rate can be estimated more easily, facilitating measurement sub-partitioning during tracking of nearby extended targets and improving the accuracy of estimating the number of targets in multi-extended target tracking. Attached Figure Description

[0037] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0038] Figure 1 This is a flowchart of the multi-extended target measurement partitioning method based on OPTICS-FCM in an embodiment of the present invention. Detailed Implementation

[0039] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0040] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0041] To address the problem of inaccurate target number estimation caused by inaccurate measurement segmentation in existing multi-extended target tracking methods, this invention proposes a multi-extended target measurement segmentation method based on OPTICS-FCM clustering. This method can segment measurement samples of varying densities generated by multiple extended targets in dense sea clutter scenarios. It not only reduces the sensitivity to input parameters but also optimizes the OPTICS clustering method, enabling it to output not only cluster structures but also clutter measurements. Based on the needs of multi-extended target tracking, it outputs the number of extended target measurement samples to be segmented, as well as the sample subsets of each segment after each clustering. Furthermore, it estimates the measurement generation rate and uses this estimated rate as the initial cluster number data, which is then input into the FCM algorithm to complete the measurement sub-segmentation.

[0042] The existing OPTICS clustering analysis method is an improved algorithm of DBSCAN clustering. It is insensitive to input parameters; as long as the value of MinPts is determined, slight changes in the radius eps will not affect the clustering results. OPTICS does not explicitly generate clusters, but instead generates an augmented cluster order (e.g., a coordinate graph with reachability distance as the vertical axis and the output order of sample points as the horizontal axis). This order represents the density-based clustering structure of each sample point. From this order, the clustering results of the DBSCAN algorithm based on any parameters eps and MinPts can be obtained, overcoming the shortcomings of using a set of global parameters in cluster analysis and effectively solving the problem of poor clustering results caused by different densities.

[0043] Based on the measurement data characteristics of radar multi-extended target tracking, this invention improves the OPTICS method. Instead of setting the initial RD value of each sample point to a maximum value, it sets it to RDTH based on the maximum Euclidean distance between the measurement sample points. Furthermore, the estimated measurement rate is introduced into FCM during the measurement subdivision stage, which effectively eliminates clutter and solves the problem of inaccurate measurement subdivision quantity.

[0044] like Figure 1 As shown in the figure, a multi-extended target measurement partitioning method based on OPTICS-FCM clustering in an embodiment of the present invention includes the following steps:

[0045] S1. Obtain the multi-extended target measurement set Z at time k. k , for Z k Preprocessing is performed, the Euclidean distance between each pair of measured samples is calculated, and the maximum value is denoted as RDTH.

[0046] Preprocessing the measurement data generates operable data processing conditions that can adaptively adjust the condition settings, thereby improving the accuracy of measurement segmentation.

[0047] S2. Set the parameter MinPts, which is the minimum number of samples in the neighborhood of the measurement sample that becomes the core point.

[0048] This parameter only plays an auxiliary role and is generally set to 5, so it has little impact on the overall effect.

[0049] S3. Input the measurement data, RDTH and MinPts into the OPTICS clustering algorithm. Use RDTH as the initial reachability distance for all measurement samples. Finally, an ordered output result and the corresponding reachability distances can be obtained, namely the order sequence and the reachability distance RD table for each measurement sample.

[0050] Specifically, S3 includes the following steps:

[0051] S31. Calculate the pairwise distances of all measured sample points. If the number of sample points |N(x)| within the neighborhood radius is greater than or equal to MinPts, then this point is called a core point. Calculate the core distance cd(x) of the core point. In the neighborhood of the current core point o, the sample point ranked in ascending order of distance to core point o is d(x,N) of the distance to o. MinPts (x) is the core distance cd(x) of o (the actual number of sample points contained in o's neighborhood is MinPts+1).

[0052]

[0053] S32. Create two queues: the pending queue QP and the result queue order;

[0054] The queue to be processed is used to store samples in the neighborhood of the core sample and their reachability distance, and is sorted in ascending order of reachability distance; the result queue is used to store the output order of sample points, which is the data that has been processed.

[0055] S33, If the target measurement set Z is expanded further. kThe algorithm terminates if all points have been processed or no core points exist. Otherwise, select an unprocessed sample point o that is a core point, first add o to the result queue order, and then remove o from QP; then find the measurement set Z. k All density values ​​of o are directly connected to sample point x. Calculate the reachability distance from x to o. If x is not in the queue QP, put x and its reachability distance into QP. If x is in QP, update the reachability distance of x if the new reachability distance of x is smaller. Finally, reorder the data in QP according to the reachability distance from smallest to largest.

[0056]

[0057] S34. If the queue QP is empty, return to step S33; otherwise, take the first sample point y (the sample point with the smallest reach) from QP and put it into the order, then add it to the measurement set Z. k The marking has been processed. The marking method is to mark the corresponding position of the sequence to be processed at position o as empty, and at the same time push o into the result queue order.

[0058] S35. If y is not a core point, repeat step S34, that is, find the sample point with the smallest reachable distance of the remaining data in QP; if y is a core point, find all density reachable sample points of y in the measurement set, calculate the reachable distance to y, and then update all density reachable sample points in QP according to S33.

[0059] S36. Repeat steps S33 and S34 until all measurement samples have been processed and the iteration terminates when the waiting queue QP is emptied. Finally, an ordered output result and the corresponding reachable distance, i.e., the order and RD table, can be obtained.

[0060] S4. Traverse the RD sequence and perform threshold judgment on the RD data.

[0061] Step S3 outputs the reachability distance sequence RD and the order sequence, sets the number of subsets count to 0, iterates through the RD sequence, and performs a threshold judgment on RD(order(i)). If RD(order(i)) is not less than RDTH and RD(order(i+1)) is less than RDTH, then the count value is increased by 1, and the sample output corresponding to the RD value before the next RD value that meets the condition is the measurement partition subset Zp. k (1), Zpk(1)=Z k (order(i)) After the RD traversal is completed, output each measurement subset Zpk(p) and the number of subsets count. The number of subsets is the estimated number of expansion targets. There are no clutter samples in the measurement subset samples, and the clutter removal is completed.

[0062] If RD(order(i)) is not less than RDTH and RD(order(i+1)) is less than RDTH, then let Z be the order of RDTH. k The i-th sample is used as the dividing point. The number of grooves in the RD data is found based on this point. This number is the number of subsets after measurement division. Samples that do not meet the conditions are marked as clutter and are not processed. Measurement samples that meet the conditions are stored in the groove data. After the RD traversal is completed, each measurement subset Zpk(p) is output. The number of subsets is the estimated number of expansion targets. There are no clutter samples in the measurement subset samples, and the clutter removal is completed.

[0063] Step S4 can remove measurement data marked as clutter, which on the one hand prevents erroneous triggering of measurement subdivision, and on the other hand avoids tracking errors caused by excessively large measurement box division in the tracking method based on box particle filtering technology.

[0064] S5, Target Measurement Rate L k The measurement rate is estimated by using the average number of samples in each subset as the measurement rate estimate.

[0065] Since clutter removal has been completed for each measurement subset, the average number of samples in each subset can be used as the measurement rate estimate based on the relationship between the number of samples in each subset. The number of samples in each measurement subset is Zp. k (p) is the number of columns to complete the measurement rate estimation.

[0066] After removing clutter, the measurement generation rate can be estimated more easily, which facilitates measurement sub-division when tracking nearby extended targets and improves the accuracy of estimating the number of targets when tracking multiple extended targets.

[0067] S6. Perform sub-partitioning judgment on each measurement subset. If the measurement sub-partitioning conditions are met, use FCM to perform measurement sub-partitioning, and then output the final measurement partition subset Zp. k (p).

[0068] The measurement rate L is estimated based on time k. k The number of measurements in each measurement subset is judged. If the number of measurement data columns in a measurement subset is greater than 1.5L, then... k If the measurement subset is composed of intersecting or adjacent targets, then FCM measurement sub-partitioning is performed on the measurement subset, and the final measurement sub-partitioning result is output, that is, the final measurement subsets Zp. k (p).

[0069] Corresponding to the multi-extended target measurement partitioning method based on OPTICS-FCM clustering in this application, this application also provides a multi-extended target measurement partitioning device based on OPTICS-FCM clustering, comprising:

[0070] The preprocessing unit is used to acquire a set of multiple extended target measurements at a preset time, preprocess the set of multiple extended target measurements, calculate the distance between pairs of measurement samples, and obtain the maximum distance.

[0071] The parameter setting unit is used to set the minimum number of samples in the neighborhood where the measurement sample becomes the core point.

[0072] The OPTICS clustering unit is used to input the measurement data, maximum distance, and minimum number of neighborhood samples obtained by the parameter setting unit from the multi-extended target measurement set obtained by the preprocessing unit into the OPTICS clustering algorithm. The maximum distance is used as the initial reachable distance of all measurement samples to obtain an ordered output result and the corresponding reachable distance.

[0073] The clutter removal unit is used to traverse the reachability sequence of the measurement samples obtained by the OPTICS clustering unit, perform threshold judgment on the reachability data of the measurement samples, and complete clutter removal.

[0074] The measurement rate estimation unit is used to estimate the measurement rate by taking the average number of samples in each subset after the clutter removal unit removes clutter as the measurement rate estimate.

[0075] The FCM partitioning unit is used to perform sub-partitioning judgment on each measurement subset based on the measurement rate estimation obtained by the measurement rate estimation unit. If the measurement sub-partitioning conditions are met, FCM is used for measurement sub-partitioning, and then the final measurement partitioning subset is output.

[0076] For the multi-extended target measurement partitioning device based on OPTICS-FCM clustering in this embodiment of the invention, since it corresponds to the multi-extended target measurement partitioning method based on OPTICS-FCM clustering in the above embodiment, the description is relatively simple. For related similarities, please refer to the description of the multi-extended target measurement partitioning method based on OPTICS-FCM clustering in the above embodiment, which will not be described in detail here.

[0077] This application also discloses a computer-readable storage medium storing a computer instruction set. When the computer instruction set is executed by a processor, it implements a multi-extended target measurement partitioning method based on OPTICS-FCM clustering as provided in any of the above embodiments.

[0078] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A multi-extended target measurement partitioning method based on OPTICS-FCM clustering, characterized in that, Includes the following steps: Obtain a multi-extended target measurement set at a preset time, preprocess the multi-extended target measurement set, calculate the distance between each pair of measurement samples, and obtain the maximum distance. Set the minimum number of samples in the neighborhood of the core point to be the measurement sample; The measurement data, maximum distance, and minimum number of neighborhood samples in the multi-extended target measurement set are input into the OPTICS clustering algorithm. The maximum distance is used as the initial reachable distance for all measurement samples to obtain an ordered output result and the corresponding reachable distance. Traverse the reachable distance sequence of the measurement samples, perform threshold judgment on the reachable distance data of the measurement samples, and complete the clutter removal; The measurement rate estimation is completed by using the average number of samples in each subset as the measurement rate estimate. Based on the measurement rate estimation, each measurement subset is subdivided. If the measurement subdivision conditions are met, FCM is used to subdivide the measurement subsets. Finally, the final measurement subdivision subset is output. The measurement sub-division conditions include: the measurement subset consists of intersecting or adjacent targets.

2. The multi-extended target measurement partitioning method based on OPTICS-FCM clustering according to claim 1, characterized in that, The measurement data, maximum distance, and minimum number of neighborhood samples in the multi-extended target measurement set are input into the OPTICS clustering algorithm. The maximum distance is used as the initial reachability distance for all measurement samples, resulting in an ordered output and corresponding reachability distances, including: Step 1: Calculate the core distance of the core point, current core point. In the neighborhood of the core point The sample point whose distance from the smallest neighbor in ascending order is... The distance as Core distance; Step 2: Create two queues, a queue to be processed. and result queue The queue to be processed stores samples in the neighborhood of the core sample and their reachability distances, and is sorted in ascending order of reachability distance; the result queue stores the output order of the sample points, which is the data that has been processed. Step 3: If all points in the multi-expanded target measurement set have been processed or no core point exists, the algorithm ends; otherwise, select an unprocessed sample point that is a core point. First of all Add to the result queue In, and from Delete Then find the measurement set. middle All densities reach the sample point ,calculate arrive The reachable distance, if Not in the pending queue In the middle, then and its reachable distance In the middle, if exist If If the new reachable distance is smaller, then update. The reachable distance, and finally for The data in the middle is reordered from smallest to largest by reachability; Step 4, if the queue to be processed If empty, return to step 3; otherwise, retrieve. The first sample point , put In the measurement set The marking has been processed; the marking method is by marking the object. The position corresponding to the position to be processed in the sequence is marked as empty, and at the same time... Pushed into the results queue middle; Step 5, if If it's not the core point, repeat step 4, i.e., find... The remaining data can reach the sample point with the smallest distance; if If it's the core point, then find it. All density points in the measurement set are directly connected to the sample points, and the values ​​are calculated to... The reachable distance is then determined, and then all density direct-reach sample points are updated according to step 3. middle; Step 6: Repeat steps 3 and 4 until all measurement samples have been processed and are in the processing queue. When the data is cleared, the iteration terminates; finally, an ordered output result and the corresponding reachable distance are obtained.

3. The multi-extended target measurement partitioning method based on OPTICS-FCM clustering according to claim 1, characterized in that, Traverse the reachability sequence of the measurement samples and perform threshold judgment on the reachability data of the measurement samples, including: Set the number of subsets The initial value is 0, and the distance that can be reached by traversing the measurement samples is... Sequence, pair Perform threshold judgment, if Not less than and Less than ,but The value increases by 1, and in the next condition that is met... Value before The corresponding sample output is a measurement partition subset. , , After the traversal is complete, output each measurement subset. and the number of subsets The number of subsets is the estimated number of expanded targets. The measurement subset samples no longer contain clutter samples, thus completing clutter removal.

4. The multi-extended target measurement partitioning method based on OPTICS-FCM clustering according to claim 1, characterized in that, Based on the aforementioned measurement rate estimation, each measurement subset is subdivided. Those subsets meeting the subdivision criteria are then subdivided using FCM, including: Based on the estimated measurement rate, the number of measurements in each measurement subset is determined. If the number of measurement data columns in a measurement subset is greater than 1.5... Then, FCM measurement subdivision is performed on the measurement subset. This represents the estimated measurement rate at time k.

5. The multi-extended target measurement partitioning method based on OPTICS-FCM clustering according to claim 1, characterized in that, The distance mentioned is the Euclidean distance.

6. The multi-extended target measurement partitioning method based on OPTICS-FCM clustering according to claim 1, characterized in that, The minimum number of samples in the neighborhood is set to 5.

7. A multi-extended target measurement and segmentation device based on OPTICS-FCM clustering, characterized in that, include: The preprocessing unit is used to acquire a set of multiple extended target measurements at a preset time, preprocess the set of multiple extended target measurements, calculate the distance between pairs of measurement samples, and obtain the maximum distance. The parameter setting unit is used to set the minimum number of samples in the neighborhood where the measurement sample becomes the core point. The OPTICS clustering unit is used to input the measurement data, maximum distance, and minimum number of neighborhood samples obtained by the parameter setting unit from the multi-extended target measurement set obtained by the preprocessing unit into the OPTICS clustering algorithm. The maximum distance is used as the initial reachable distance of all measurement samples to obtain an ordered output result and the corresponding reachable distance. The clutter removal unit is used to traverse the reachability sequence of the measurement samples obtained by the OPTICS clustering unit, perform threshold judgment on the reachability data of the measurement samples, and complete clutter removal. The measurement rate estimation unit is used to estimate the measurement rate by taking the average number of samples in each subset after the clutter removal unit removes clutter as the measurement rate estimate. The FCM partitioning unit is used to perform sub-partitioning judgment on each measurement subset based on the measurement rate estimation obtained by the measurement rate estimation unit. If the measurement sub-partitioning conditions are met, FCM is used for measurement sub-partitioning, and then the final measurement partitioning subset is output.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer instruction set, which, when executed by a processor, implements a multi-extended target measurement partitioning method based on OPTICS-FCM clustering as described in any one of claims 1 to 6.

Images