A radar photoelectric signal processing method for multi-target detection

Abstract

The application discloses a radar photoelectric signal processing method for multi-target detection and belongs to the field of radar multi-target identification. The method solves the problem of poor fusion identification effect of radar signals and photoelectric signals in the multi-target identification process. The technical scheme comprises the following steps: synchronously acquiring an original intermediate frequency signal sequence of a phased array radar and a laser reflection spectrum image sequence of a laser radar, and extracting radar observation data vectors and laser observation data vectors; aligning the laser field of view and the radar field of view, and realizing a three-dimensional common field of view; calculating mutual information values of the radar observation data and the laser observation data; calculating an adaptive adjustment factor, jointly filtering the radar signals, and outputting a radar signal frequency domain response; and calculating the correlation probability between targets, and evaluating whether the targets are the same real target through the correlation probability. The application proposes a probability data correlation model, which is used for more accurately acquiring the number of targets and target trajectories in the field of view and realizing high-precision multi-target detection.

Description

Technical Field

[0001] This invention relates to the field of radar multi-target recognition, and more specifically to a radar photoelectric signal processing method for multi-target detection. Background Technology

[0002] Phased array radar and lidar are two different types of radar systems. Phased array radar uses radio waves, or electromagnetic waves, as its signal source. It measures range, velocity, and azimuth by transmitting and receiving electromagnetic waves. Lidar uses light waves, or lasers, as its signal source. Lidar obtains the range, azimuth, and velocity information of a target by emitting a laser beam and receiving the reflected signal. Phased array radar changes the beam direction by altering the phase of the radar wave, belonging to electronically scanned radar. It utilizes a dense antenna array, with each antenna capable of transmitting and receiving radar waves. During scanning, one or more blocks can be selected for scanning. Phased array radar requires no mechanical rotation, offering high flexibility and making it suitable for dealing with highly maneuverable targets. Lidar, on the other hand, uses lasers as its emission source and performs active remote sensing through photoelectric detection technology. Lidar scanning methods mainly include mechanical and semi-solid-state methods (such as MEMS and micro-mirrors). The advantages of lidar lie in its high resolution and high precision, but its performance is poor in rainy or foggy conditions.

[0003] In traditional multi-target detection, radar and optoelectronic systems often operate independently or undergo simple data-layer fusion. Radar has ranging and velocity measurement capabilities, but low angular resolution and susceptibility to clutter interference. Optoelectronic systems (such as lidar) have high angular resolution and strong feature recognition capabilities, but are greatly affected by weather and lack direct ranging capabilities. Fusing the two systems during target detection can mutually optimize their weaknesses. Existing fusion methods are mostly fixed-weight fusion or simple logical decision-making, which suffer from high association error rates, missed detection of weak targets, ambiguous target attribute discrimination, and insufficient real-time performance in environments with dense targets, cross-maneuvering targets, and strong interference. Summary of the Invention

[0004] To address the aforementioned shortcomings of existing technologies, this invention provides a radar photoelectric signal processing method for multi-target detection, which significantly improves the detection, tracking accuracy, and recognition accuracy in complex multi-target scenarios.

[0005] To achieve the above-mentioned objectives, the technical solution adopted by this invention is as follows:

[0006] A radar photoelectric signal processing method for multi-target detection is provided, comprising:

[0007] Step S1: Simultaneously acquire the original intermediate frequency signal sequence of the phased array radar and the laser reflection spectrum image sequence of the lidar, extract the observation sequence of the radar signal and the photoelectric observation sequence of the laser signal, and obtain the radar observation data vector and the laser observation data vector.

[0008] Step S2: Align the laser field of view and the radar field of view, and divide the field of view into several three-dimensional grids based on the azimuth angle, elevation angle and observation distance latitude to obtain a three-dimensional common field of view; assign a virtual observation distance to the target observed by the laser signal, and calculate the mutual information value between the radar observation data and the laser observation data based on the number of times the target observed by the radar signal and the target observed by the laser signal fall on different three-dimensional grids.

[0009] Step S3: Calculate the adaptive adjustment factor using the mutual information value, perform joint filtering on the radar signal, and output the frequency domain response of the jointly filtered radar signal;

[0010] Step S4: Based on the radar signal after joint filtering, acquire the radar signal feature vector of the target, and based on the laser observation data vector, acquire the laser signal feature vector of the target. Calculate the correlation probability between the target observed by the radar and the target observed by the laser, and evaluate whether they are the same real target through the correlation probability.

[0011] Further, step S1 includes:

[0012] Step S11: Synchronously acquire the raw intermediate frequency signal sequence of the phased array radar. Laser reflection spectrum image sequence of lidar And perform spatiotemporal alignment. t For time; estimate the original intermediate frequency signal sequence Signal power spectral density and radar noise power spectral density , f The frequency of the radar signal;

[0013] Step S12: Using the original intermediate frequency signal sequence Constructing observation sequences of radar signals , For radar observation data vectors, , i The target number observed by radar signals. The target i Range, azimuth, and elevation angles within the radar field of view The number of targets observed by the radar signal is T, where T is the transpose operation;

[0014] Using laser reflectance spectral image sequences Constructing photoelectric observation sequences of laser signals , This is a vector of laser observation data. ; j The target number observed by the laser signal. The number of targets observed by the laser signal. Respectively as target j Azimuth angle, elevation angle, laser intensity and spectral feature vector in laser field of view.

[0015] Further, step S2 comprises:

[0016] Step S21: field of view alignment is performed on the laser field of view and the radar field of view to obtain a common field of view, and the common field of view is divided into grid cells in the azimuth angle and elevation angle dimensions, and the radar observation range is divided into L intervals to form a three-dimensional common field of view divided into three-dimensional grid cells;

[0017] Step S22: define the three-dimensional grid coordinates on the three-dimensional common field of view, obtain the number of times a target falls into different three-dimensional grid cells according to the radar observation data vector i , assign a virtual observation distance to the target based on the laser observation data vector j , and obtain the number of times a target falls into different three-dimensional grid cells; j

[0018] Step S23: count the number of times j a target and a target i fall into a three-dimensional grid cell at the same time , calculate the radar edge probability , the photoelectric edge probability and the joint probability ; use the radar edge probability , the photoelectric edge probability and the joint probability to calculate the mutual information value of the radar observation data and the laser observation data.

[0019] Further, step S3 comprises:

[0020] Step S31: use the mutual information value to calculate the adaptive adjustment factor ;

[0021] The mutual information value is used to measure the information complementarity between the radar signal and the laser signal, and the smaller the mutual information value, the more the laser signal can assist the radar signal to retain more radar signal details.

[0022] Step S32: combine the adaptive adjustment factor ​​Jointly filter the radar signals, and output a frequency domain response of the jointly filtered radar signals .

[0023] Further, the step S4 comprises:

[0024] Step S41: Collect a target radar signal feature vector based on the jointly filtered radar signals i , , a cross-section fluctuation feature of the jointly filtered radar signals, respectively a distance, an azimuth angle and a pitch angle of the target i in the radar field of view; obtain a target laser signal feature vector based on the laser observation data vector j , , a texture feature of the laser signal of the target j ;

[0025] Step S42: fuse the radar signal feature vector and the laser signal feature vector into a shared latent space vector ;

[0026] Step S43: calculate a correlation probability of the target i and the target j using the latent space vector ;

[0027] Step S44: set a correlation probability threshold , if , determine that the target i and the target j are the same real target, if , determine that the target i and the target j are not the same real target, and two different real targets appear in the radar field of view and the laser field of view.

[0028] A terminal device is provided, which comprises a processor, a transceiver and a memory, the memory is used to store a computer program, the processor is used to call and run the computer program from the memory, and control the transceiver to perform receiving or sending actions, so that the terminal device performs the above-mentioned radar-optical signal processing method for multi-target detection.

[0029] A computer storage medium is provided, which is used to store a computer program, and the computer program comprises instructions for executing the above-mentioned radar-optical signal processing method for multi-target detection.

[0030] The beneficial effects of the present application are: the present application filters radar signals and photoelectric signals jointly by mutual information entropy guidance, optimizes radar signals, and improves the probability of weak target detection in strong clutter and low signal-to-noise ratio environment, while reducing false alarm rate. The present application also proposes a probability data association model, which can effectively handle the situation of target dense intersection and similar features. Compared with the traditional nearest neighbor association, the correct rate of association between targets is effectively improved, especially when the photoelectric part is partially shielded or the radar is flickering; the number of targets in the field of view and the target trajectory can be more accurately obtained, and high-precision multi-target detection is realized. BRIEF DESCRIPTION OF DRAWINGS

[0031] Figure 1 A flowchart of a radar photoelectric signal processing method for multi-target detection. DETAILED DESCRIPTION

[0032] The specific embodiments of the present application are described below to facilitate understanding of the present application by those skilled in the art, but it should be clear that the present application is not limited to the scope of the specific embodiments. It is obvious to those skilled in the art that any changes within the spirit and scope of the present application defined and determined by the appended claims are obvious, and all inventions utilizing the concept of the present application are within the scope of protection.

[0033] As shown in Figure 1 , a radar photoelectric signal processing method for multi-target detection, comprising:

[0034] Step S1: synchronously acquiring a raw intermediate frequency signal sequence of a phased array radar and a laser reflection spectrum image sequence of a laser radar, extracting an observation sequence of the radar signal and a photoelectric observation sequence of the laser signal, and obtaining a radar observation data vector and a laser observation data vector. Step S1 specifically comprises:

[0035] Step S11: synchronously acquiring a raw intermediate frequency signal sequence of a phased array radar and a laser reflection spectrum image sequence of a laser radar , and performing time and space alignment, t is time; estimating the signal power spectral density in the raw intermediate frequency signal sequence and the radar noise power spectral density , f is the frequency of the radar signal;

[0036] Step S12: constructing an observation sequence of the radar signal using the raw intermediate frequency signal sequence , is a radar observation data vector, , i is the target number observed by the radar signal,​ respectively are the target i in the radar field of view in the distance, azimuth and elevation angles, is the number of targets observed by the radar signal;

[0037] using the laser reflection spectrum image sequence constructing the photoelectric observation sequence of the laser signal , is the laser observation data vector, ; j is the target number observed by the laser signal, is the number of targets observed by the laser signal, respectively are the target j in the laser field of view in the azimuth, elevation angle, laser intensity and spectrum feature vector.

[0038] Step S2: Align the laser field of view and the radar field of view, and divide the field of view into a plurality of three-dimensional grids based on the azimuth, elevation angle and observation distance latitude, to obtain a three-dimensional common field of view; a virtual observation distance is assigned to the target observed by the laser signal, and the mutual information value of the radar observation data and the laser observation data is calculated according to the number of times the target observed by the radar signal and the target observed by the laser signal fall in different three-dimensional grids. Step S2 specifically comprises:

[0039] Step S21: Align the laser field of view and the radar field of view to obtain a common field of view, and divide the common field of view into grids in the azimuth and elevation angle dimensions, and simultaneously divide the radar observation distance range into L intervals, forming a three-dimensional common field of view divided into three-dimensional grids;

[0040] Step S22: define the three-dimensional grid coordinates on the three-dimensional common field of view, obtain the number of times the target falls in different three-dimensional grids according to the radar observation data vector i , assign a virtual observation distance to the target based on the laser observation data vector j , and obtain the number of times the target falls in different three-dimensional grids. j ;

[0041] ;

[0042] wherein, respectively are the target set observed by the radar signal, the target set observed by the laser signal, is the historical time observation window​ inside and target j target with the closest flight speed i distance within the radar field of view, respectively, the target speed observed by the radar signal, the target speed observed by the laser signal, is the distance correction coefficient;

[0043] distance correction coefficient According to the difference between the identification accuracy of the laser radar to the target and the identification accuracy of the phased array radar to the target, when the identification accuracy of the laser radar to the target is greater than the identification accuracy of the phased array radar to the target, the distance correction coefficient is taken in the range of 1-2, otherwise, the distance correction coefficient is taken in the range of 0-1;

[0044] Step S23: Count the target j and target i fall into a three-dimensional grid at the same time , calculate the radar edge probability , the photoelectric edge probability and the joint probability ; use the radar edge probability , the photoelectric edge probability and the joint probability to calculate the mutual information value of the radar observation data and the laser observation data ;

[0045] ;

[0046] wherein, is the total smoothing coefficient, is the smoothing coefficient in three directions in the three-dimensional common field of view, and generally the total smoothing coefficient and the smoothing coefficient are taken as very small positive numbers (such as ), to avoid the case of .

[0047] Step S3: Calculate the adaptive adjustment factor using the mutual information value, jointly filter the radar signal, and output the radar signal frequency domain response after joint filtering. Step S3 specifically includes:

[0048] Step S31: Calculate the adaptive adjustment factor using the mutual information value ;

[0049] ;

[0050] wherein, is the standard deviation of the mutual information value based on different targets;

[0051] Mutual information value The smaller the mutual information value is, the more details of the radar signal can be reserved by the laser signal.

[0052] Step S32: combining the adaptive adjustment factor Jointly filtering the radar signal to output a joint filtered radar signal frequency domain response

[0053]

[0054] Step S4: acquiring a radar signal feature vector of the target based on the joint filtered radar signal, acquiring a laser signal feature vector of the target based on the laser observation data vector, calculating a correlation probability between the target observed by the radar and the target observed by the laser, and evaluating whether they are the same real target through the correlation probability. Step S4 specifically comprises:

[0055] Step S41: acquiring a radar signal feature vector of the target based on the joint filtered radar signal i , is a cross section fluctuation feature of the joint filtered radar signal, are the distance, azimuth angle and elevation angle of the joint filtered target i in the radar field of view; acquiring a laser signal feature vector of the target j , is a texture feature of the laser signal of the target j ;

[0056] Step S42: fusing the radar signal feature vector and the laser signal feature vector into a shared latent space vector ;

[0057] Step S43: calculating a correlation probability of the target i and the target j using the latent space vector ;

[0058] ;

[0059] wherein, are the feature likelihood coefficients and the spatio-temporal feature correlation coefficient of the target i and the target j , is a false alarm likelihood item, ​​​​is a potential space clustering center, is a Mahalanobis distance weight matrix, is a target i is a target j is a consistency probability of azimuth in a laser field of view and a radar field of view, is a target i is a target j is a covariance matrix of azimuth in a laser field of view and a radar field of view, is a target i is a target j is a consistency probability of distance in a laser field of view and a radar field of view, is a target i is a target j is a variance of distance in a laser field of view and a radar field of view, is a probability density function;

[0060] correlation probability The confidence degree that a phased array radar observation process and a laser radar observation process correspond to a same real target is quantitatively calculated, and the size of the correlation probability can effectively represent the probability that two targets i , target j are a same real target, the greater the correlation probability is, the greater the probability that the target i , target j is a same real target is.

[0061] Step S44: set a correlation probability threshold , if , it is determined that the target i and the target j are a same real target, if , it is determined that the target i and the target j are not a same real target, and two different real targets appear in the radar field of view and the laser field of view.

[0062] The present application sticks together the kinematics information of the radar and the characteristic information of the photoelectricity by calculating the correlation probability, forms a more complete description of the target, the laser signal assists the radar signal to identify and monitor the target in the field of view, and the number of targets and target trajectories in the field of view can be more accurately acquired, and high-precision multi-target detection is realized; the radar photoelectric system can make more stable and smoother decisions under uncertainty.

[0063] If the association probability of a radar observation target with all existing targets is low, and the association probability with all photoelectric observations is also generally low, it is likely to be a starting point of a new target or a false alarm, and the radar photoelectric system can start the new track starting logic accordingly. If a plurality of frames of an existing target track cannot be associated with the laser observation process with high probability, it can be determined that the target has left or disappeared.

[0064] A terminal device comprises a processor, a transceiver and a memory, the memory is used for storing a computer program, the processor is used for calling and running the computer program from the memory, and the transceiver is used for executing receiving or sending actions, so that the terminal device executes the radar photoelectric signal processing method for multi-target detection.

[0065] A computer storage medium is used for storing a computer program, and the computer program comprises instructions for executing the radar photoelectric signal processing method for multi-target detection.

[0066] The present application jointly filters radar signals and photoelectric signals by mutual information entropy, optimizes radar signals, improves the probability of weak target detection in a strong clutter and low signal-to-noise ratio environment, and reduces the false alarm rate. The present application also proposes a probability data association model, which can effectively handle the situation of target dense intersection and similar features, effectively improves the association accuracy between targets compared with the traditional nearest neighbor association, and is stable especially when the photoelectric part is shielded or the radar is flickering; the number of targets in the field of view and the target track can be more accurately obtained, and high-precision multi-target detection is realized.

Claims

1. A radar photoelectric signal processing method for multi-target detection, characterized in that, include: Step S1: Simultaneously acquire the original intermediate frequency signal sequence of the phased array radar and the laser reflection spectrum image sequence of the lidar, extract the observation sequence of the radar signal and the photoelectric observation sequence of the laser signal, and obtain the radar observation data vector and the laser observation data vector. Step S2: Align the laser field of view and the radar field of view, and divide the field of view into several three-dimensional grids based on the azimuth angle, elevation angle and observation distance latitude to obtain a three-dimensional common field of view; assign a virtual observation distance to the target observed by the laser signal, and calculate the mutual information value between the radar observation data and the laser observation data based on the number of times the target observed by the radar signal and the target observed by the laser signal fall on different three-dimensional grids. Step S3: Calculate the adaptive adjustment factor using the mutual information value, perform joint filtering on the radar signal, and output the frequency domain response of the jointly filtered radar signal; Step S4: Based on the jointly filtered radar signal, acquire the radar signal feature vector of the target; based on the laser observation data vector, acquire the laser signal feature vector of the target; calculate the correlation probability between the target observed by radar and the target observed by laser; and evaluate whether they are the same real target by using the correlation probability.

2. The radar photoelectric signal processing method for multi-target detection according to claim 1, characterized in that, Step S1 includes: Step S11: Synchronously acquire the raw intermediate frequency signal sequence of the phased array radar. Laser reflection spectrum image sequence of lidar And perform spatiotemporal alignment. t For time; estimate the original intermediate frequency signal sequence Signal power spectral density and radar noise power spectral density , f The frequency of the radar signal; Step S12: Using the original intermediate frequency signal sequence Constructing observation sequences of radar signals , For radar observation data vectors, , i The target number observed by radar signals. The target i Range, azimuth, and elevation angles within the radar field of view The number of targets detected by radar signals; Using laser reflectance spectral image sequences Constructing photoelectric observation sequences of laser signals , This is a vector of laser observation data. ; j The target number observed by the laser signal. The number of targets observed by the laser signal. The target j Azimuth, elevation, laser intensity, and spectral eigenvectors within the laser field of view.

3. The radar photoelectric signal processing method for multi-target detection according to claim 2, characterized in that, Step S2 includes: Step S21: Align the laser field of view and the radar field of view to obtain a common field of view, and divide the common field of view into azimuth and elevation dimensions. Each grid simultaneously divides the radar observation range into [number] grids. L The intervals are divided into several sections. A shared three-dimensional field of view of a three-dimensional grid; Step S22: Define the coordinates of the stereo mesh on the common stereo field of view. According to radar observation data vector Obtain the target i Number of times falling on different 3D grids Based on laser observation data vector Towards the target j Assign a virtual observation distance and obtain the target j Number of times falling on different 3D grids ; Step S23: Statistical Objective j and target i Number of times simultaneously falling into a 3D grid Calculate radar edge probability Optoelectronic edge probability and joint probability ; Utilizing radar edge probability Optoelectronic edge probability and joint probability Calculate the mutual information value between radar observation data and laser observation data. .

4. The radar photoelectric signal processing method for multi-target detection according to claim 3, characterized in that, Step S3 includes: Step S31: Utilize mutual information values Calculate the adaptive adjustment factor ; Step S32: Combine with adaptive adjustment factor Perform joint filtering on the radar signal and output the frequency domain response of the jointly filtered radar signal. .

5. The radar photoelectric signal processing method for multi-target detection according to claim 4, characterized in that, Step S4 includes: Step S41: Acquire target based on radar signal after joint filtering i Radar signal feature vector , The characteristics of radar signal cross-sectional area fluctuations after joint filtering. The target after joint filtering i Range, azimuth, and elevation angles within the radar field of view; target acquisition based on laser observation data vectors. j laser signal feature vector , For the goal j Texture features of the laser signal; Step S42: Convert the radar signal feature vector With laser signal feature vector Merge into a shared latent space vector ; Step S43: Utilize latent space vectors Calculation target i With the goal j Association probability ; Step S44: Set the association probability threshold ,like Then the target is determined. i With the goal j For the same real goal, if Then the target is determined. i With the goal j They are not the same real target; two different real targets appeared in the radar field of view and the laser field of view.

6. A terminal device, characterized in that, The device includes a processor, a transceiver, and a memory. The memory stores a computer program, and the processor retrieves and runs the computer program from the memory. The transceiver is controlled to perform receiving or transmitting actions, thereby enabling the terminal device to execute the radar photoelectric signal processing method for multi-target detection as described in any one of claims 1-5.

7. A computer storage medium, characterized in that, Used to store computer programs, the computer programs including instructions for executing the radar photoelectric signal processing method for multi-target detection as described in any one of claims 1-5.

Images