Multi-human body target separation method based on multi-dimensional parameter estimation

Through MIMO radar and virtual antenna expansion technology, combined with multiple signal processing methods, the problem of target distinction and high computational complexity in multi-object detection is solved, and high-precision and high-efficiency multi-body target separation is achieved.

CN118914982APending Publication Date: 2024-11-08无锡芯湖湾信息科技有限公司
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410951911.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-16
Publication Date
2024-11-08

AI Technical Summary

Technical Problem

When existing radar signal processing technology detects and separates multiple targets, especially when the target distance is close or the same, it is difficult to accurately distinguish targets. The calculation complexity of the 2D-MUSIC algorithm is high, which limits real-time processing capabilities and widespread applications. .

Method used

Using MIMO radar technology and time division multiplexing mode, the angle resolution is improved through virtual antenna expansion technology, combined with 2D-FFT, SVD noise reduction processing, median filtering and binary processing, to form a sub-search space, using one-dimensional MUSIC algorithm and The cascading MUSIC algorithm reduces the computational complexity and achieves the target two-dimensional parameter estimation.

Benefits of technology

It significantly improves the estimation accuracy of target angle information, reduces the calculation amount of 2D-MUSIC algorithm, improves the accuracy and real-timeness of target detection, and can accurately estimate angle and distance parameters in multiple human target scenarios, suitable for fast response real-time multi-object detection.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118914982A_ABST
    Figure CN118914982A_ABST
Patent Text Reader

Abstract

The invention provides a multi-dimensional parameter estimation-based multi-human body target separation method, which comprises the following steps of: constructing an antenna array by using an MIMO radar technology, transmitting a frequency modulation signal by adopting a time division multiplexing mode, and improving the angular resolution by using a virtual antenna expansion technology; performing 2D-FFT processing on radar receiving signals, generating a distance angle matrix RAM, and realizing preliminary estimation of angles and distances of a plurality of targets; singular value decomposition (SVD) noise reduction processing is carried out on the distance angle matrix RAM, noise is removed through median filtering, and target signal features are enhanced; performing binarization processing on the distance angle matrix RAM after noise reduction, and determining an initial distance and angle estimation interval of the target; estimating the angle by adopting a one-dimensional MUSIC algorithm to obtain an angle estimation value of the target; and the angle estimation value is substituted into a 2D-MUSIC algorithm, accurate estimation of the distance is carried out, and a two-dimensional parameter estimation value of the target is obtained. And the target identification capability and the target separation resolution are enhanced.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of signal processing, and more specifically, to a method for separating multiple human targets based on multi-dimensional parameter estimation. Background Art

[0002] In the field of radar signal processing, multi-target detection and separation is an important but challenging task. When multiple targets are located at the same or similar distance, traditional detection methods often have difficulty distinguishing these targets. In addition, multi-target detection based on CFAR shows limitations when the targets are located in the same distance unit, and it is necessary to introduce additional angular information dimensions to achieve effective separation.

[0003] As an algorithm that can simultaneously estimate the target direction and distance, the 2D-MUSIC algorithm has theoretical advantages, but its computational complexity is high, especially when it involves high-dimensional covariance matrix operations and two-dimensional spatial spectrum peak search. This not only affects the real-time processing capability of the algorithm, but also limits its widespread use in practical applications. Summary of the invention

[0004] The present invention provides a method for separating multiple human targets based on multidimensional parameter estimation, which is characterized by comprising the following steps:

[0005] Step 1: Use MIMO radar technology to build an antenna array, use time division multiplexing mode to transmit frequency modulation signals, and improve the angular resolution through virtual antenna expansion technology;

[0006] Step 2: Perform 2D-FFT processing on the radar receiving signal to generate a range angle matrix RAM to achieve preliminary estimation of the angles and distances of multiple targets;

[0007] Step 3: Perform singular value decomposition (SVD) noise reduction processing on the distance angle matrix RAM, and remove noise through median filtering to enhance the target signal characteristics;

[0008] Step 4: Binarize the denoised distance angle matrix RAM to determine the preliminary distance and angle estimation interval of the target;

[0009] Step 5: Based on the binarization result, the distance angle matrix RAM map is segmented to form a sub-search space of the target, thereby reducing the computational complexity of the spectrum peak search;

[0010] Step 6: Use a one-dimensional MUSIC algorithm to estimate the angle and obtain an estimated angle value of the target;

[0011] Step 7: Substitute the angle estimation value into the 2D-MUSIC algorithm to accurately estimate the distance and obtain the two-dimensional parameter estimation value of the target.

[0012] In some embodiments, the MIMO radar described in step 1 adopts a 2×4 configuration, including two physical transmitting antennas and four physical receiving array elements, and four virtual receiving antennas obtained by expanding the virtual technology.

[0013] In some embodiments, the 2D-FFT in step 2 includes Range-FFT and Angle-FFT, and the distance information of the target is obtained through Range-FFT, and the angle information of the target is obtained through Angle-FFT.

[0014] In some embodiments, the SVD denoising process described in step 3 selects the main singular values ​​of the distance angle matrix RAM for denoising, and the median filtering algorithm uses a square window with a window size of 2N+1, where N is the width of the filter window edge.

[0015] In some embodiments, the binarization process described in step 4 adopts a threshold determination method, and the binarization threshold is selected as the boundary between the target signal and the background noise in the segmentation distance angle matrix RAM map.

[0016] In some embodiments, the segmented distance angle matrix RAM map described in step 5 is based on the binarization result, and the area with a value of 1 in the distance angle matrix RAM map is used as the estimated range of the target signal to form a sub-search space.

[0017] In some embodiments, the one-dimensional MUSIC algorithm described in step 6 utilizes the spectrum peak characteristics of the signal to perform accurate angle estimation within the initially estimated angle interval.

[0018] In some embodiments, the 2D-MUSIC algorithm described in step 7 combines the accurate estimated value of the angle to perform accurate distance estimation within the distance estimation interval, thereby achieving accurate estimation of the two-dimensional parameters of the target.

[0019] In some embodiments, the method further includes computational complexity analysis, optimizing the decomposition of the covariance matrix by a Nystrom algorithm to reduce computational complexity, and further reducing the computational complexity of the spectrum peak search by a cascaded MUSIC algorithm.

[0020] In some embodiments, the method further comprises simplifying the 2D parameter estimation problem into a one-dimensional estimation problem by using a cascaded MUSIC algorithm, thereby achieving accurate estimation of the two-dimensional parameters of the target.

[0021] The above embodiments of the present invention have at least the following beneficial effects:

[0022] Utilizing MIMO radar technology and time division multiplexing mode, the virtual antenna expansion technology significantly improves the angular resolution and enhances the estimation accuracy of target angle information.

[0023] The Nystrom algorithm is used to optimize the decomposition of the covariance matrix, which effectively reduces the computational complexity of the 2D-MUSIC algorithm and improves the efficiency of subspace decomposition.

[0024] The cascade MUSIC algorithm is used to simplify the 2D parameter estimation problem into a one-dimensional estimation problem, which greatly reduces the computational complexity of the spectrum peak search and improves the practicality and efficiency of the algorithm.

[0025] Through SVD noise reduction processing and median filtering, noise is effectively removed and target signal characteristics are enhanced, improving the accuracy of target detection and signal recognition capabilities;

[0026] It achieves effective separation of multiple human targets, especially when the targets are close or at the same distance, and can accurately estimate the angle and distance parameters of each target;

[0027] Through simulation experiments and actual data verification tests, it is proved that the method of the present invention has good performance and high adaptability in different scenarios;

[0028] The method of the present invention significantly improves the running speed of the algorithm while ensuring the accuracy of parameter estimation, and is suitable for real-time multi-target detection scenarios that require rapid response. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] The above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily understood by reading the detailed description below with reference to the accompanying drawings. In the accompanying drawings, several embodiments of the present invention are shown in an exemplary and non-limiting manner, in which:

[0030] Figure 1 A schematic flow chart of a method for separating multiple human targets based on multi-dimensional parameter estimation provided by one embodiment of the present invention;

[0031] Figure 2 A schematic diagram of a flow chart of a secondary MUSIC parameter estimation algorithm based on partitioned RAM provided by an embodiment of the present invention;

[0032] Figure 3 A schematic diagram of constructing a distance angle matrix provided by an embodiment of the present invention;

[0033] FIG4( a ) is a schematic diagram of a RAM after 2D-FFT processing provided by an embodiment of the present invention;

[0034] FIG4( b ) is a schematic diagram of a RAM after SVD noise reduction according to an embodiment of the present invention;

[0035] FIG4( c ) is a schematic diagram of a RAM after median filtering provided by an embodiment of the present invention;

[0036] FIG4( d ) is a schematic diagram of a binarized RAM according to an embodiment of the present invention;

[0037] FIG4( e) is a schematic diagram of RAM partitioning provided by an embodiment of the present invention;

[0038] FIG. 4( f ) is an enlarged view of RAM partitioning provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0039] The principles and spirit of the present invention will be described below with reference to several exemplary embodiments. It should be understood that these embodiments are provided only to enable those skilled in the art to better understand and implement the present invention, and are not intended to limit the scope of the present invention in any way. On the contrary, these embodiments are provided to make the present invention more thorough and complete, and to fully convey the scope of the present invention to those skilled in the art.

[0040] Those skilled in the art know that the embodiments of the present invention can be implemented as a system, device, apparatus, method or computer program product. Therefore, the present invention can be specifically implemented in the following forms, namely: complete hardware, complete software (including firmware, resident software, microcode, etc.), or a combination of hardware and software.

[0041] It should be noted that any number of elements in the drawings is for illustration rather than limitation, and any naming is only for distinction and does not have any limiting meaning.

[0042] Reference below Figure 1 , Figure 1 A schematic flow chart of a method for separating multiple human targets based on multidimensional parameter estimation provided by one embodiment of the present invention. A method for separating multiple human targets based on multidimensional parameter estimation provided by one embodiment of the present invention includes:

[0043] Step 1: Use MIMO radar technology to build an antenna array, use time division multiplexing mode to transmit frequency modulation signals, and improve the angular resolution through virtual antenna expansion technology;

[0044] In some embodiments, MIMO (Multiple-Input Multiple-Output) radar technology is a method of using multiple transmit and receive antennas to improve radar system performance. It enhances radar detection and resolution of targets through spatial diversity and signal processing techniques.

[0045] In some embodiments, the MIMO radar adopts a 2×4 configuration, including two physical transmitting antennas and four physical receiving array elements, and four virtual receiving antennas obtained by expanding the virtual technology.

[0046] The 2×4 configuration refers to a radar system with two physical transmit antennas and four physical receive array elements. This configuration allows the radar to transmit and receive signals from multiple directions at the same time, increasing the system's flexibility and target detection capabilities.

[0047] Time Division Multiplexing (TDM) is a signal transmission method that allows multiple signals to use the same communication medium in different time periods. In radar systems, TDM allows the transmitting antenna to transmit signals alternately in different time periods, thereby improving the transmission efficiency of the system.

[0048] It should be noted that it is a method of creating additional receiving antennas at the software level through signal processing technology, thereby increasing the number of receiving channels of the system and improving angular resolution. Through MIMO technology and virtual antenna expansion, the radar system can determine the direction of the target more accurately, which is especially important for distinguishing multiple targets, thereby improving angular resolution.

[0049] Step 2: Perform 2D-FFT processing on the radar receiving signal to generate a range angle matrix RAM to achieve preliminary estimation of the angles and distances of multiple targets;

[0050] 2D-FFT (Two-Dimensional Fast Fourier Transform) is a mathematical transformation method used to convert a signal from a time (or space) domain to a frequency domain. In the present invention, 2D-FFT processing specifically refers to performing FFT on a signal in two dimensions, first a Range-FFT in the distance dimension, and then an Angle-FFT in the angle dimension.

[0051] RAM (Range-Angle Matrix) is a distance angle matrix, which is a matrix composed of 2D-FFT processing results. Its rows correspond to different distance units, and its columns correspond to different angle units. The elements in the matrix represent the signal strength at the corresponding distance and angle units.

[0052] In some embodiments, generating the distance angle matrix RAM may include the following steps:

[0053] First, Range-FFT is performed on the received radar signal at each sampling point of the sweep frequency, which generates a range-frequency matrix with each row representing a range unit and each column representing a frequency component.

[0054] Next, the results of the Range-FFT are stored by column, and angle estimation data is prepared for each range unit.

[0055] Then, an Angle-FFT is performed on these data, which will generate an angle-frequency matrix with each row representing an angle unit and each column representing a distance unit.

[0056] Finally, by accumulating the angle information of multiple chirps, the angle-frequency matrix is ​​converted into a distance-angle matrix RAM, in which each element represents the signal strength at a specific distance and angle.

[0057] Preliminary estimates of the angles and distances of multiple targets can include: Using the peak position of the signal strength in RAM, the distance and angle of the target can be preliminarily estimated. The peak of the signal strength usually corresponds to the actual position of the target. By analyzing the distribution of the signal strength in RAM, the existence of multiple targets can be identified and preliminary angle and distance estimates can be made for them. The results of the preliminary estimate provide a basis for subsequent precise estimates, especially when the targets are close or overlapped. This step is particularly important.

[0058] Step 3: Perform singular value decomposition (SVD) noise reduction processing on the distance angle matrix RAM, and remove noise through median filtering to enhance the target signal characteristics;

[0059] It should be noted that SVD (Singular Value Decomposition) is a mathematical method used to decompose a matrix into three specific matrix products. In signal processing, SVD can be used to identify and remove noise components. Singular values ​​are part of the SVD decomposition result and represent the importance of the signal. The main singular values ​​correspond to the main components of the signal, while smaller singular values ​​are usually associated with noise.

[0060] Median filtering is a nonlinear signal processing technique that reduces noise by replacing the value of each point with the median value of its neighborhood. This method removes noise while maintaining the edges of the signal.

[0061] In some embodiments, the SVD denoising process selects the main singular values ​​of the distance angle matrix RAM for denoising, and the median filtering algorithm uses a square window with a window size of 2N+1, where N is the width of the filter window edge.

[0062] The window size is 2N+1, which is the size of the median filter, where N is the filter window edge width. Larger windows can remove noise more effectively, but may blur rapid changes in the signal.

[0063] Step 4: Binarize the denoised distance angle matrix RAM to determine the preliminary distance and angle estimation interval of the target;

[0064] Binarization is an image processing technique that converts pixels in an image into only two possible gray levels, usually black and white, to highlight the boundary between the target and the background. Threshold determination method An automatic method for determining the binarization threshold, such as the Otsu method, selects the optimal threshold based on the histogram of the image to maximize the difference between the target and the background.

[0065] In some embodiments, the binarization process adopts a threshold determination method, and the binarization threshold is selected as the boundary between the target signal and the background noise in the segmentation distance angle matrix RAM diagram.

[0066] Step 5: Based on the binarization result, the distance angle matrix RAM map is segmented to form a sub-search space of the target, thereby reducing the computational complexity of the spectrum peak search;

[0067] The RAM image can be segmented to divide the pixels in the RAM image into target signals and background noise based on the binarization result, thus forming a sub-search space of the target.

[0068] In some embodiments, the segmented distance angle matrix RAM map is based on a binarization result, and the area with a value of 1 in the distance angle matrix RAM map is used as an estimated range of the target signal to form a sub-search space.

[0069] The sub-search space is a region in the RAM graph that contains the target signal with a preliminary estimate. By limiting the search range, the amount of computation in subsequent processing steps can be reduced.

[0070] Step 6: Use a one-dimensional MUSIC algorithm to estimate the angle and obtain an estimated angle value of the target;

[0071] In some embodiments, the one-dimensional MUSIC algorithm uses the spectrum peak characteristics of the signal to perform accurate angle estimation within a preliminarily estimated angle interval.

[0072] The one-dimensional MUSIC algorithm (Multiple Signal Classification) is a high-resolution spectrum estimation technique used to estimate the direction of a signal. One-dimensional MUSIC here refers to executing the MUSIC algorithm in only one dimension (angle or distance). The MUSIC algorithm is used to accurately estimate the target in the angle dimension and determine the direction of the target by analyzing the characteristics of the signal subspace.

[0073] Step 7: Substitute the angle estimation value into the 2D-MUSIC algorithm to accurately estimate the distance and obtain the two-dimensional parameter estimation value of the target.

[0074] In some embodiments, the 2D-MUSIC algorithm combines the accurate estimated value of the angle to accurately estimate the distance within the distance estimation interval, thereby achieving accurate estimation of the two-dimensional parameters of the target.

[0075] The 2D-MUSIC algorithm is a two-dimensional version of the MUSIC algorithm, which can simultaneously estimate signals in two dimensions: angle and distance. Combined with the accurate estimated value of the angle, the distance is accurately estimated within the distance estimation interval, achieving accurate estimation of the two-dimensional parameters of the target.

[0076] In some embodiments, the method further includes computational complexity analysis, optimizing the decomposition of the covariance matrix by a Nystrom algorithm to reduce computational complexity, and further reducing the computational complexity of the spectrum peak search by a cascaded MUSIC algorithm.

[0077] The Nystrom algorithm is a method for approximating the eigenvalues ​​and eigenvectors of large matrices by selecting a subset of the matrix to reduce the computational complexity. In signal processing, the covariance matrix describes the statistical characteristics of the signal. The Nystrom algorithm optimizes this decomposition process and improves efficiency.

[0078] The method also includes simplifying the 2D parameter estimation problem into a one-dimensional estimation problem through a cascade MUSIC algorithm, thereby achieving accurate estimation of the two-dimensional parameters of the target.

[0079] In some embodiments, the cascaded MUSIC algorithm is a method that combines the one-dimensional and two-dimensional MUSIC algorithms, first accurately estimates the angle dimension, and then uses the result for two-dimensional distance estimation. By first accurately estimating the angle and then using it for distance estimation, the complexity of the problem is simplified, the amount of calculation is reduced, and the accuracy of the estimation is maintained.

[0080] In some embodiments, the core of the 2D-MUSIC algorithm is to use the orthogonal properties of the steering vector of the array signal and the noise subspace to construct a spectrum peak search function and realize parameter estimation through grid search. In order to estimate the distance and angle parameters at the same time, it is necessary to reconstruct the steering vector on the basis of maintaining orthogonality so that it integrates the distance and angle information to construct a two-dimensional spectrum peak search function. Its expression is as follows:

[0081]

[0082] In the formula, E n represents the noise subspace obtained by eigenvalue decomposition, is the reconstructed steering vector, where eT(θ) and eR(R) are:

[0083]

[0084] From formula (1), we can see that the K maximum peaks of P(θ, R) correspond to the angle and distance estimation values ​​of the target. The computational complexity of the 2D MUSIC algorithm is mainly concentrated in the two-dimensional space search and the decomposition of the covariance matrix, and the computational complexity is O(M 2 L+q[(MK)(M+1)]) and O(M 3 ), where M is the number of array elements, L is the number of snapshots, and q is the number of grids. Obviously, with the increase of the number of snapshots and the density of grids, the amount of calculation will increase exponentially, which seriously restricts the real-time processing capability of the system.

[0085] In order to solve the problem of high computational complexity of 2D-MUSIC, this section introduces a Nystrom algorithm to accelerate the decomposition of subspaces. This algorithm effectively reduces the computational complexity by selecting a subset of the matrix to approximate the eigenvalues ​​and eigenvectors of the entire matrix.

[0086] First, decompose the covariance matrix R into:

[0087]

[0088] In the formula, K≤z≤NQ.

[0089] Through the Nystrom function, only R 11 and R 21 The noise subspace of the covariance matrix R can be approximated by the information of the received data X:

[0090]

[0091] In the formula,

[0092] Therefore, the autocorrelation of X1 and X2 is calculated separately, and we get:

[0093]

[0094] Then for R 11 Perform eigenvalue decomposition and get:

[0095]

[0096] Using the Nystrom algorithm extension, we can get an approximate characteristic matrix:

[0097]

[0098] In the above formula, the matrix The mutual orthogonality of column vectors is not satisfied, so orthogonalization is required. For GH G performs eigenvalue decomposition:

[0099]

[0100] Finally, we get the approximate characteristic matrix that satisfies orthogonality:

[0101]

[0102] Approximate noise subspace E n Consists of the last MQ-K columns of matrix E.

[0103] Although the 2D-MUSIC algorithm based on Nystrom solves the computational complexity of the matrix decomposition algorithm, the computational complexity of the 2D-MUSIC algorithm is mainly concentrated in the spectrum peak search process. Especially in the environment of pursuing high precision, the improvement of grid accuracy makes the computational complexity of the 2D-MUSIC algorithm increase exponentially. Therefore, in response to the above problems, this paper proposes a cascade MUSIC algorithm based on RAM estimation, which can be divided into two main stages: preliminary estimation and fine estimation. Figure 2 As shown in Figure 1, the preliminary estimation stage involves four steps: first, a 2D-FFT is performed on the input signal to generate a RAM; then, the RA matrix is ​​denoised using the singular value decomposition (SVD) technique; then, a median filter algorithm is applied to remove sharp noise in the signal; finally, the filtered signal is binarized and the RAM is segmented to obtain the initial distance and angle estimation intervals of each target.

[0104] In the fine estimation stage, the Nystrom algorithm is introduced to improve the efficiency of subspace decomposition and reduce the computational complexity. Secondly, in the process of spectrum peak search, a cascade MUSIC algorithm is proposed, which firstly makes a one-dimensional MUSIC precise estimation of the angle, and then substitutes this precise angle estimation value into 2D-MUSIC to infer the precise value of the distance. This algorithm successfully reduces the computational complexity of the algorithm from two-dimensional multiplication-level operations to one-dimensional addition-level operations, significantly improving the overall estimation efficiency.

[0105] In some embodiments, the received signal is first subjected to a first FFT at each frequency sweep sampling point to obtain the distance information of the target, i.e., Range-FFT. The data processed by Range-FFT is stored in a data buffer column by column. Subsequently, a second FFT is performed in the spatial dimension to obtain the angle information of the target, i.e., Angle-FFT. Finally, the distance angle matrix is ​​constructed by accumulating the angle information of multiple Chirps in the data frame. The specific process is as follows: Figure 3As shown in the figure. The 2D-FFT algorithm can estimate the angles of multiple human targets and separate them by constructing RAM. However, when performing multiple human target detection in a real environment, the target chest angle estimated by 2D-FFT deviates from the actual angle. This is because the human target in reality is not an ideal point target, but has a certain width.

[0106] SVD decomposition and median filtering are used to perform noise reduction on the obtained distance angle matrix, and finally the matrix is ​​binarized.

[0107] The SVD denoising process can be expressed as:

[0108] R ra =U*Λ*V H ,R ra ∈A N*Q (10)

[0109] In the formula, R ra Represents the distance angle matrix, U=[u1,u2,...,u n ],U ra ∈A N*N and V = [v1,v 2, ...,v q ],V ra ∈A Q*Q are N-order and Q-order unitary matrices, respectively, whose column vectors are left and right singular vectors, and are mutually orthogonal; Λ∈B N*Q is a diagonal matrix whose off-diagonal elements are all 0, and the elements on the diagonal are singular values, and are arranged in descending order. i ={λ1>λ2>...>λ K}.

[0110] Expanding formula (10), it can be expressed as:

[0111]

[0112] Since the first few singular values ​​have covered the main energy, this paper selects the first k singular values ​​to perform denoising and reconstruction on the distance angle matrix, where k is the smallest integer that satisfies the following conditions:

[0113]

[0114] The reconstructed distance angle matrix is ​​expressed as follows:

[0115]

[0116] The median filter technique performs well in removing noise points in the radar range angle map and can retain the structure of the target signal, especially in retaining edge information, which is superior to other filtering methods, which is of great significance for target recognition and classification. In RAM processing, the target appears as a bright spot or bright spot. The median filter enhances the performance of the target feature by smoothing the surrounding random noise, making the target signal more significant, which is conducive to further target detection and information extraction. The original matrix is ​​padded before the median filter so that the median filter can also be applied to the edge. The formula for the square median filter is:

[0117]

[0118] Wherein, n=1,2,...,N, q=1,2,...,Q, and marg is the width of the filter window edge, and its value is set to 1.

[0119] After median filtering, the distance and angle are binarized to obtain a preliminary estimate of the distance and angle. It can be expressed as:

[0120]

[0121] In the formula, the binary threshold T selection formula is:

[0122]

[0123] After binarization, the distance estimation range and angle estimation range are obtained. The RAM graph is divided to obtain sub-search spaces.

[0124] Determine the distance estimation range range_bin of each target k based on the binarization result k and the angle estimation range angle_bin k , forming the corresponding sub-search space.

[0125] FIG4 shows the preliminary estimation results of the distance angle of the two targets. In FIG4(a), 2D-FFT is used to analyze the received signal, and the preliminary separation of the position and angle of the detected target is successfully achieved. FIG4(b) shows the result after denoising by SVD, which effectively removes the Gaussian noise in the distance angle matrix. In FIG4(c), the image is processed by median filtering technology to more clearly show the detection range. FIG4(d) is an image after binarization processing. The area with a value of 1 clearly marks the possible distance and angle estimation range of the detected target. FIG4(e) and FIG4(f) are schematic diagrams of RAM segmentation. By segmenting the RAM graph, the search range of distance and angle is significantly reduced, thereby reducing the amount of calculation in the spectrum peak search process and improving the calculation efficiency.

[0126] In some examples, based on the angle interval obtained by the preliminary estimation, the MUSIC algorithm is independently executed for each target to obtain an accurate estimate of the target angle. Subsequently, the obtained accurate angle value is substituted into the 2D-MUSIC algorithm and calculated within the distance estimation interval to obtain an accurate estimate of the corresponding distance.

[0127] Table 1 shows the comparison of the computational complexity of the three 2D-MUSIC methods. In this table, M represents the number of array antenna elements, K represents the number of signal sources, L represents the number of snapshots, q represents the total number of grids in the 2D-MUSIC search space, R (r1, r2...r k ),A(a1,a2…a k ) represent the number of search cycles in the distance and angle dimensions of each sub-search space obtained after segmenting the RAM graph, and Z represents the block parameter selected by the Nystrom algorithm. The computational complexity of the spectrum peak search of the 2D-MUSIC algorithm is O p {q[M(MK)+MK]}, the computational complexity of the spectrum peak search based on the cascade MUSIC algorithm of partitioned RAM is O p_级联MUSIC {(R+A)[M(MK)+MK]}.

[0128]

[0129] Table 1 Comparison of the amount of computation required by different algorithms

[0130] right Figure 3The results in the above analysis are analyzed. The number of array antennas is 8, the number of signal sources is 2, the number of snapshots is 100, the Nystrom algorithm block parameter is 2, the distance dimension spectrum peak search range is [0,10], the search step is 0.1m, the angle dimension spectrum peak search range is [-90°,90°], and the search step is 0.1°. The total number of grids in the 2D-MUSIC search space is 1800, and the sub-search spaces are [11.6:19.6,1.8:2.2], [-9.4:-16.2,4.8:5.1], [23. 8:32.8,0.8:1.0]. After analysis, the computational complexity of the three algorithms are (6400+512+1800(48+8-2))=104112, (1600+32+1800(48+8-2))=98832, (1600+32+(80+4+68+3+90+2)(48+8-2))=14970. Obviously, the computational complexity of the 2D-MUSIC algorithm based on Nystrom has been significantly improved in the eigenvalue decomposition process. As the number of array elements and snapshots increases, the advantages of the algorithm become more prominent. However, the main computational complexity of the two-dimensional parameter estimation is still concentrated in the spectrum peak search process. From the above analysis, we can see that the RAM-based cascade MUSIC algorithm is only 1 / 7 of the traditional 2D-MUSIC, and the complexity is reduced by 7 times; and when the grid accuracy reaches 0.01, the RAM-based cascade MUSIC algorithm is only 1 / 72 of the traditional 2D-MUSIC, and the complexity is reduced by 72 times. Therefore, the cascade MUSIC algorithm based on split RAM has a great advantage in computational complexity, and with the improvement of grid accuracy, its computational effect will be more significant.

[0131] The multi-human target separation method based on multi-dimensional parameter estimation of the present invention significantly improves the angular resolution of target detection by combining MIMO radar technology and time division multiplexing mode, and enhances the target recognition ability through virtual antenna expansion technology. The distance angle matrix RAM generated by 2D-FFT provides an effective means for preliminarily estimating the angles and distances of multiple targets. The application of SVD noise reduction and median filtering technology effectively removes the noise in the signal, enhances the target signal characteristics, and improves the accuracy of signal recognition. The covariance matrix decomposition optimized by the Nystrom algorithm and the application of the cascaded MUSIC algorithm greatly reduce the computational complexity and improve the operation efficiency and real-time performance of the algorithm. In addition, the method shows good robustness in a variety of complex environments, and its effectiveness and applicability are verified by simulation experiments and actual data tests, providing a practical, efficient and accurate solution for real-time multi-target detection.

[0132] The electronic devices in some embodiments of the present disclosure may include but are not limited to mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc.

[0133] The electronic device may include a processing device (such as a central processing unit, a graphics processor, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) or a program loaded from a storage device to a random access memory (RAM). In the RAM, various programs and data required for the operation of the electronic device are also stored. The processing device, ROM, and RAM are connected to each other through a bus. An input / output (I / O) interface is also connected to the bus.

[0134] Typically, the following devices may be connected to the I / O interface: input devices including, for example, a touch screen, a touch pad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, etc.; output devices including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; storage devices including, for example, a magnetic tape, a hard disk, etc.; and communication devices. The communication devices may allow the electronic device to communicate with other devices wirelessly or by wire to exchange data.

[0135] In particular, according to some embodiments of the present disclosure, the process described above with reference to the flowchart can be implemented as a computer software program. For example, some embodiments of the present disclosure include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program contains program code for executing the method shown in the flowchart. In some such embodiments, the computer program can be downloaded and installed from a network through a communication device, or installed from a storage device, or installed from a ROM. When the computer program is executed by a processing device, the above-mentioned functions defined in the method of some embodiments of the present disclosure are executed.

[0136] It should be noted that some embodiments of the present disclosure may also include a computer-readable medium, and the computer-readable storage medium may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In some embodiments of the present disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program, which may be used by or in combination with an instruction execution system, device or device. In some embodiments of the present disclosure, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, in which a computer-readable program code is carried. This propagated data signal may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. The computer readable signal medium may also be any computer readable medium other than a computer readable storage medium, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, apparatus or device. The program code contained on the computer readable medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination of the above.

[0137] In some embodiments, the client and the server may communicate using any currently known or future developed network protocol such as HTTP (HyperText Transfer Protocol), and may be interconnected with any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network ("LAN"), a wide area network ("WAN"), an internet (e.g., the Internet), and a peer-to-peer network (e.g., an ad hoc peer-to-peer network), as well as any currently known or future developed network.

[0138] The above-mentioned computer-readable medium may be included in the above-mentioned electronic device; or it may exist separately without being assembled into the electronic device. The computer program code for performing the operations of some embodiments of the present disclosure may be written in one or more programming languages ​​or a combination thereof, and the above-mentioned programming languages ​​include object-oriented programming languages, such as Java, Smalltalk, C++, and also include conventional procedural programming languages, such as "C" language or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as an independent software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., using an Internet service provider to connect through the Internet).

[0139] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present disclosure. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some implementations as replacements, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0140] The units described in some embodiments of the present disclosure may be implemented in software or in hardware. The units described may also be provided in a processor, and the functions described above herein may be performed at least in part by one or more hardware logic components. For example, and without limitation, exemplary types of hardware logic components that may be used include: field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), complex programmable logic devices (CPLDs), and the like.

[0141] The above description is only some preferred embodiments of the present disclosure and an explanation of the technical principles used. Those skilled in the art should understand that the scope of the invention involved in the embodiments of the present disclosure is not limited to the technical solutions formed by the specific combination of the above-mentioned technical features, but should also cover other technical solutions formed by any combination of the above-mentioned technical features or their equivalent features without departing from the above-mentioned inventive concept. For example, the above-mentioned features are replaced with the technical features with similar functions disclosed in the embodiments of the present disclosure (but not limited to) to form a technical solution. The above description is only some preferred embodiments of the present invention and an explanation of the technical principles used. Those skilled in the art should understand that the scope of the invention involved in the embodiments of the present invention is not limited to the technical solutions formed by the specific combination of the above-mentioned technical features, but should also cover other technical solutions formed by any combination of the above-mentioned technical features or their equivalent features without departing from the above-mentioned inventive concept. For example, the above-mentioned features are replaced with the technical features with similar functions disclosed in the embodiments of the present invention (but not limited to) to form a technical solution.

[0142] The above descriptions are only some preferred embodiments of the present invention and an explanation of the technical principles used. Those skilled in the art should understand that the scope of the invention involved in the embodiments of the present invention is not limited to the technical solutions formed by a specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above inventive concept. For example, the above features are replaced with (but not limited to) technical features with similar functions disclosed in the embodiments of the present invention to form a technical solution.

Claims

1. A method for separating multiple human targets based on multidimensional parameter estimation, characterized in that: The following steps are involved: Step 1: Use MIMO radar technology to build an antenna array, use time division multiplexing mode to transmit frequency modulation signals, and improve the angular resolution through virtual antenna expansion technology; Step 2: Perform 2D-FFT processing on the radar receiving signal to generate a range angle matrix RAM to achieve preliminary estimation of the angles and distances of multiple targets; Step 3: Perform singular value decomposition (SVD) noise reduction processing on the distance angle matrix RAM, and remove noise through median filtering to enhance the target signal characteristics; Step 4: Binarize the denoised distance angle matrix RAM to determine the preliminary distance and angle estimation interval of the target; Step 5: Based on the binarization result, the distance angle matrix RAM map is segmented to form a sub-search space of the target, thereby reducing the computational complexity of the spectrum peak search; Step 6: Use a one-dimensional MUSIC algorithm to estimate the angle and obtain an estimated angle value of the target; Step 7: Substitute the angle estimation value into the 2D-MUSIC algorithm to accurately estimate the distance and obtain the two-dimensional parameter estimation value of the target.

2. The method according to claim 1, characterized in that The MIMO radar described in step 1 adopts a 2×4 configuration, including two physical transmitting antennas and four physical receiving array elements, as well as four virtual receiving antennas obtained through virtual technology expansion.

3. The method according to claim 1, characterized in that The 2D-FFT described in step 2 includes Range-FFT and Angle-FFT. The distance information of the target is obtained by Range-FFT, and the angle information of the target is obtained by Angle-FFT.

4. The method according to claim 1, characterized in that: The SVD denoising process described in step 3 selects the main singular values ​​of the distance angle matrix RAM for denoising, and the median filtering algorithm uses a square window with a window size of 2N+1, where N is the width of the filter window edge.

5. The method according to claim 1, characterized in that The binarization process described in step 4 adopts a threshold determination method, and the binarization threshold is selected as the boundary between the target signal and the background noise in the segmentation distance angle matrix RAM map.

6. The method according to claim 1, characterized in that The segmented distance angle matrix RAM map described in step 5 is based on the binarization result, and the area with a value of 1 in the distance angle matrix RAM map is used as the estimated range of the target signal to form a sub-search space.

7. The method according to claim 1, characterized in that The one-dimensional MUSIC algorithm described in step 6 utilizes the spectrum peak characteristics of the signal to perform accurate angle estimation within the initially estimated angle interval.

8. The method according to claim 1, characterized in that The 2D-MUSIC algorithm described in step 7 combines the accurate estimated value of the angle to accurately estimate the distance within the distance estimation interval, thereby achieving accurate estimation of the two-dimensional parameters of the target.

9. The method according to claim 1, characterized in that: The method also includes a computational complexity analysis, optimizing the decomposition of the covariance matrix by a Nystrom algorithm to reduce the computational complexity, and further reducing the computational complexity of the spectrum peak search by a cascaded MUSIC algorithm.

10. The method according to claim 1, characterized in that The method also includes simplifying the 2D parameter estimation problem into a one-dimensional estimation problem through a cascade MUSIC algorithm, thereby achieving accurate estimation of the two-dimensional parameters of the target.