Life target detection method, device and equipment using stepped frequency continuous wave through-the-wall radar

Abstract

The application provides a life target detection method, device and equipment using a stepped frequency continuous wave through-wall radar. In a time-frequency domain level, pulse compression based on an inverse fast Fourier transform is first performed on received multi-channel echo signals to obtain a high-resolution one-dimensional range image. Then, fast Fourier transform is performed along a slow time dimension to realize coherent accumulation, a range-Doppler matrix is constructed, a harmonic product spectrum technology is applied to multi-scale spectrum fusion of the range-Doppler matrix to obtain a range-frequency domain matrix. In a space domain level, a multi-channel space-time-frequency correlation matrix is constructed based on the range-frequency domain matrix, automatic identification and distance pairing of a target are realized in combination with a constant false alarm rate detection, and finally, a two-dimensional spatial coordinate of the target is solved by using an elliptical intersection positioning method. In this way, the problem of high false alarm and missed alarm and poor robustness under a low signal-to-noise ratio condition is effectively solved, and weak life target detection of the through-wall radar is realized.

Description

Technical Field

[0001] This application relates to the field of data processing technology, and in particular to a method, apparatus and equipment for detecting living targets using stepped-frequency continuous-wave through-wall radar. Background Technology

[0002] For complex scenarios such as post-disaster rubble search and rescue, target reconnaissance under typical wall obstruction, and counter-terrorism operations, radar signals need to penetrate consumable media such as brick walls, concrete, or rubble to detect concealed life targets.

[0003] Stepped Frequency Continuous Wave (SFCW) through-wall radar is a core technique in non-contact life detection. The basic principle of SFCW radar-based life detection is based on the Doppler effect. When a person is at rest, breathing and heartbeat cause minute movements (millimeter-level displacements) on the body surface. These movements modulate the phase (or time delay) of the radar echo. The radar receiver receives the echo signal and extracts information about the phase or amplitude fluctuations over time at the target distance, thereby enabling the detection of vital signs such as respiratory rate.

[0004] Through-wall radar target detection is one of the core technologies of through-wall radar. In practical application scenarios, due to the complex and changeable on-site environment, severe electromagnetic interference, and the effects of wall attenuation and refraction on electromagnetic waves, existing through-wall radar target detection methods generally suffer from problems such as inaccurate measurement under low signal-to-noise ratio conditions and poor environmental adaptability.

[0005] Existing through-wall radar technologies for detecting weak life signs rely on processing single-dimensional signal features, failing to acquire the target's two-dimensional spatial location. In practical applications, limited by the high attenuation of the wall medium and system random noise, this single-domain processing method often fails in low signal-to-noise ratio environments: on the one hand, ranging methods relying solely on time-domain energy are highly susceptible to noise interference, leading to deviations in time-of-arrival (TOA) estimation and consequently, divergent positioning results; on the other hand, detection methods relying solely on frequency-domain peak values ​​lack spatial constraints, making it difficult to eliminate false targets. This approach prevents existing algorithms from accurately detecting and locating weak life targets in complex through-wall scenarios.

[0006] In summary, due to the extremely low signal-to-noise ratio (SNR) caused by the two-way attenuation of the medium, and the fact that random noise in the system can easily bury weak life target signals, existing through-wall radar target detection algorithms suffer from serious false alarms and missed alarms, as well as poor reliability and robustness. Summary of the Invention

[0007] This application proposes a method, apparatus, and equipment for detecting living targets using stepped-frequency continuous-wave through-wall radar, which can solve one of the problems existing in the background art.

[0008] To achieve the above objectives, this application adopts the following technical solution:

[0009] Firstly, a method for detecting living targets using stepped-frequency continuous-wave through-wall radar is provided, including:

[0010] Obtain multi-channel fast-time data of step-frequency continuous wave through-wall radar obtained from slow-time dimension observations;

[0011] Perform an inverse fast Fourier transform on the fast time data of each channel to obtain a two-dimensional echo matrix. The size of the two-dimensional echo matrix is ​​M×N, where M is the number of fast time sampling points and N is the number of slow time sampling points.

[0012] Perform a Fast Fourier Transform along the slow time dimension on each range gate of the two-dimensional echo matrix to obtain the range-Doppler matrix;

[0013] The range-Doppler matrix is ​​processed using harmonic product spectrum techniques to obtain a range-frequency domain matrix;

[0014] Based on the distance-frequency domain matrix, a multi-channel spatiotemporal-frequency correlation matrix is ​​constructed;

[0015] A two-dimensional constant false alarm rate (CFAR) method is used to traverse the multi-channel spatiotemporal-frequency correlation matrix, mark potential target regions, and obtain multi-channel target distances; and,

[0016] Based on the multi-channel target distance, the target coordinates are calculated.

[0017] Based on the above technical solution, at the time-frequency domain level, pulse compression based on inverse fast Fourier transform is first performed on the received multi-channel echo signals to obtain a high-resolution one-dimensional range image. Subsequently, fast Fourier transform is performed along the slow time dimension to achieve coherent accumulation and construct a range-Doppler matrix. Then, harmonic product spectrum technology is applied to perform multi-scale spectrum fusion on the range-Doppler matrix to obtain a range-frequency domain matrix. At the spatial domain level, the spatial gain of the multi-channel system is fully utilized, and a multi-channel spatio-temporal-frequency correlation matrix is ​​constructed based on the range-frequency domain matrix. Combined with constant false alarm rate detection, automatic target identification and range matching are achieved. Finally, the two-dimensional spatial coordinates of the target are solved using the elliptic cross-location method. In this way, the problems of high false alarm rate and missed alarm rate and poor robustness under low signal-to-noise ratio conditions are effectively solved, and the detection of weak life targets by through-wall radar is realized.

[0018] In one possible design of the first aspect, the method for detecting living targets using stepped-frequency continuous-wave through-wall radar further includes:

[0019] Before performing the Fast Fourier Transform, the mean of the slow time series corresponding to each distance gate of the two-dimensional echo matrix is ​​calculated, and the mean is subtracted from the original sequence of the two-dimensional echo matrix to filter out the zero-frequency background component.

[0020] In one possible design of the first aspect, the method for detecting living targets using stepped-frequency continuous-wave through-wall radar further includes:

[0021] A slow-time bandpass filter is applied to the two-dimensional echo matrix, with the bandpass filter range set to 0.1–2Hz.

[0022] In one possible design approach of the first aspect, the distance-Doppler matrix is ​​expressed as:

[0023] The distance-frequency domain matrix is ​​expressed as:

[0024] Where h is the distance gate, f is the frequency, and n is the slow-time sampling point. It is a two-dimensional echo matrix. Representing the distance-Doppler matrix The spectrum of the downsampled version, where m is the distance cell index. This indicates the number of harmonics being considered.

[0025] In one possible design approach of the first aspect, the multi-channel is a dual-channel, the third... The distance-frequency domain matrix of each channel is expressed as follows: , express The Enhanced spectrum of each distance unit, express The The enhanced spectrum of each distance cell, and the multi-channel spatiotemporal correlation matrix are:

[0026] Where B represents the respiratory-related frequency band, and f represents the frequency. These are the mean values ​​within their respective bands.

[0027] In one possible design approach of the first aspect, a two-dimensional constant false alarm rate (CFAR) method is used to traverse the multi-channel spatiotemporal-frequency correlation matrix and mark potential target regions, specifically including:

[0028] A two-dimensional rectangular sliding window is used to traverse the multi-channel spatiotemporal correlation matrix. The window is set with an inner protective window and an outer training window centered on the pixel being inspected. The local background mean is estimated using the pixels in the outer training window.

[0029] The decision threshold is determined using the local background mean and the preset false alarm probability; and,

[0030] The potential target region is determined by using the multi-channel spatiotemporal correlation matrix value of the detected pixel and the decision threshold.

[0031] In one possible design of the first aspect, the method for detecting living targets using stepped-frequency continuous-wave through-wall radar also includes:

[0032] After marking potential target regions, the energy value corresponding to each range gate is obtained based on the range-frequency domain matrix;

[0033] The energy values ​​are used to construct the distance-energy spectrum corresponding to each channel;

[0034] Map the coordinates of all range cells in each potential target region back to the range energy spectrum of each channel, and select the range gate corresponding to the maximum energy value in the region as the location of the candidate target; and...

[0035] In the distance-frequency domain matrix, check whether the distance gate corresponding to the maximum energy value contains a valid respiratory peak. If it exists and the frequency deviation of the two channel peaks is within the specified error range, then determine that the potential target area corresponds to a real life target, and record the distance gate corresponding to the potential target area in each channel as the distance L from each channel to the target.

[0036] In one possible design approach of the first aspect, the channels correspond to the receiving antennas, and based on the multi-channel target distance, the target coordinates (x, y) are solved as follows:

[0037] The coordinates of the transmitting antenna are: The coordinates of receiving antenna 1 are The coordinates of receiving antenna 2 are .

[0038] Secondly, a life target detection device using stepped-frequency continuous-wave through-wall radar is provided, comprising:

[0039] The acquisition unit is used to acquire multi-channel fast-time data of step-frequency continuous wave through-wall radar obtained from slow-time dimension observations.

[0040] The first transformation unit is used to perform inverse fast Fourier transform on the fast time data of each channel to obtain a two-dimensional echo matrix. The size of the two-dimensional echo matrix is ​​M×N, where M is the number of fast time sampling points and N is the number of slow time sampling points.

[0041] The second transformation unit is used to perform a fast Fourier transform on each range gate of the two-dimensional echo matrix along the slow time dimension to obtain a range-Doppler matrix;

[0042] The enhancement unit is used to perform harmonic product spectrum processing on the range-Doppler matrix to obtain a range-frequency domain matrix;

[0043] The construction unit is used to construct a multi-channel spatiotemporal correlation matrix based on the distance-frequency domain matrix;

[0044] The traversal unit is used to traverse the multi-channel spatiotemporal-frequency correlation matrix using a two-dimensional constant false alarm rate method, mark potential target regions, and obtain multi-channel target distances; and,

[0045] The solving unit is used to solve for the target coordinates based on the multi-channel target distance.

[0046] Thirdly, an electronic device is provided, comprising: a processor, and a memory coupled to the processor, the memory for storing a computer program; the processor for executing the computer program stored in the memory such that the electronic device performs the life target detection method using stepped-frequency continuous-wave through-wall radar as described in any possible implementation of the first aspect. Attached Figure Description

[0047] To more clearly illustrate the technical solutions in the embodiments of this application, the drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0048] Figure 1 This is an algorithm flowchart of the through-wall radar weak life target detection method based on space-time-frequency joint enhancement provided in the embodiments of this application;

[0049] Figure 2 This is a schematic diagram of the preprocessing results of the first set of experimental dual-channel signals provided in the embodiments of this application;

[0050] Figure 3 This is a schematic diagram of the time-domain and frequency-domain cross-correlation of the first set of experiments provided in this embodiment, as well as the processing results of this embodiment;

[0051] Figure 4 This is a schematic diagram of the preprocessing results of the second set of experimental dual-channel signals provided in the embodiments of this application;

[0052] Figure 5This is a schematic diagram of the time-domain and frequency-domain cross-correlation of the second set of experiments provided in this application embodiment, as well as the processing results of this embodiment. Detailed Implementation

[0053] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0054] It should be noted that although functional modules are divided in the device schematic diagram and a logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than the module division in the device or the order in the flowchart. The terms "first," "second," etc., in the specification, claims, and the aforementioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence.

[0055] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit this application.

[0056] like Figure 1As shown, this embodiment proposes a method for detecting weak life targets using through-wall radar based on joint space-time-frequency enhancement. In the time-frequency domain, the method first performs pulse compression based on the inverse fast fourier transform (IFFT) on the received multi-channel echo signals to obtain a high-resolution one-dimensional range profile, followed by static background removal and slow-time bandpass filtering. Subsequently, a fast fourier transform (FFT) is performed along the slow-time dimension to achieve coherent accumulation and construct a range-Doppler (RD) matrix. Based on this, the harmonic product spectrum (HPS) technique is applied to perform multi-scale spectral fusion on the RD matrix, highlighting the periodic breathing components with harmonic structures, thereby effectively suppressing non-stationary random noise and completing signal enhancement in the time-frequency dimension. In the spatial domain, the spatial gain of the multi-channel system is fully utilized, and an enhanced cross-channel spectral correlation matrix is ​​constructed based on the time-frequency enhanced spectrum to generate a life sign feature map with joint space-time-frequency enhancement, significantly improving target saliency and suppressing spatial incoherent noise. Based on this feature map, automatic target identification and range matching are achieved by combining constant false alarm rate (CFAR) detection and energy criteria. Finally, the two-dimensional spatial coordinates of the target are solved using the elliptical cross-location method. This method effectively solves the problems of high false alarm rate and missed alarm rate and poor robustness of existing methods under low signal-to-noise ratio conditions, and realizes the detection of weak life targets by through-wall radar.

[0057] In one specific embodiment of this example, the SFCW radar transmits a sequence of continuous wave signals with linearly increasing frequency within one scanning period, including... Each frequency point, with a starting frequency of The frequency step is Then the first The frequency of each frequency point is For ease of explanation, let's assume there exists a distance of... If we consider the target and disregard multipath effects, then the human target echo in each frequency sweep cycle... It can be represented as:

[0058] in, Indicates the amplitude of the life echo signal. Represents a rectangle function. Indicates the pulse duration. This indicates the echo delay.

[0059] The received echo signal consists of direct-coupled waves, wall reflections, human target echoes, and system noise. Therefore, a single-channel radar echo signal... It can be represented as follows:

[0060] in It is a direct-coupled wave signal. The wave is reflected from the wall. For human target echo, This represents system noise. Sampling at each frequency point yields the baseband frequency domain echo sequence. It can be represented as:

[0061] in and These represent the echo amplitudes of the direct-coupled wave and the echo reflected from the wall, respectively. and The fixed time delays are for direct-coupled waves and wall reflections, respectively. This represents system noise.

[0062] To obtain a high-resolution range image, the echo signal Perform the inverse Fourier transform (IFFT). Let the number of IFFT transform points be... Then the pulse compression of the first A signal from the distance gate It can be directly expressed as:

[0063] The result of the calculation This is a one-dimensional range image. The above derivation only applies to the signal processing within a single scan cycle. To extract time-varying information including respiratory motion, the radar system needs to continuously transmit. A series of SFCW signals with several cycles can be generated. By continuously observing the slow time dimension and performing IFFT processing on the fast time data of each cycle, a signal of size [size missing] can be constructed. Two-dimensional echo matrix ,in Indicates fast time sampling points, Indicates slow time sampling points. This is the channel number.

[0064] In through-wall detection scenarios, the energy of wall echoes and direct-coupled waves is typically much stronger than human respiratory signals. To highlight vital signs, slowly varying or static background data needs to be removed from the original echo data. Since respiratory motion is a quasi-periodic signal that varies with slow time, while other background information is approximately unchanging with slow time, in this embodiment, the echo matrix... The mean of each row (i.e., the slow time series corresponding to each distance gate) is calculated and subtracted from the original sequence to filter out zero-frequency background components. The expression is:

[0065] in, The first echo in the original echo matrix The index in the th The value of a slow time sampling point. This represents the total number of sampling points in the slow time dimension.

[0066] Despite the removal of static background, high-frequency noise and low-frequency drift due to hardware instability remain in the signal. The fundamental frequency of resting respiration in normal adults is typically between 0.1 and 0.6 Hz. However, considering that the micro-movements of the chest cavity caused by actual breathing are not ideal sine waves, the echo signal contains higher harmonic components. To fully preserve this harmonic information to support subsequent spectral enhancement processing, this embodiment sets the bandpass filter range in the slow time dimension to 0.1–2 Hz while suppressing out-of-band noise.

[0067] Traditional Fast Fourier Transform (FFT) can improve the signal-to-noise ratio by utilizing slow-time accumulation, but it typically focuses only on the fundamental frequency peak, often ignoring the effective energy contained in higher harmonics. Given that respiratory signals exhibit quasi-periodic characteristics and rich harmonic structures similar to audio signals in the slow-time dimension, this embodiment introduces the HPS algorithm into the field of non-contact life detection. Based on this, this embodiment constructs a frequency domain enhancement matrix for the respiratory band, significantly improving the detection performance of weak respiratory signals.

[0068] HPS, by multiplying the original spectrum point-by-point with its several downsampled versions, enables coherent energy accumulation at the fundamental frequency of highly periodic respiratory signals in the time domain, while random noise, lacking harmonic structure, is significantly suppressed in the multi-scale product. To explain this more clearly, suppose a signal with harmonic structure is represented in the frequency domain as... HPS can be represented in the spectral domain as:

[0069] in, Represents a signal The spectrum of the downsampled version This indicates the number of harmonics being considered.

[0070] In practice, the preprocessed echo matrix is ​​first processed... Each distance gate performs an FFT along the slow-time dimension to accumulate energy in the respiratory signal, significantly improving the signal-to-noise ratio. Subsequently, a distance-Doppler matrix is ​​constructed. This completes the transformation from the time domain to the frequency domain. Let the first... The frequency domain signal of each distance gate is The calculation formula is as follows:

[0071] Subsequently, HPS processing is performed on the matrix, and the enhanced range-frequency domain matrix can be expressed as:

[0072] in, Representing a matrix The spectrum of the downsampled version This indicates the number of harmonics being considered.

[0073] HPS effectively suppresses uncorrelated random noise, improves the correlation of cross-channel signals, and enhances the detectability of vital signs under low signal-to-noise ratio conditions, providing a reliable foundation for subsequent weak target extraction based on cross-correlation analysis. However, relying solely on single-channel time-frequency enhancement is insufficient to completely eliminate noise interference and fails to acquire target location information. Therefore, this embodiment further introduces spatial dimension information, utilizing the spatial correlation of respiratory signals from the same human target in different spatial channels to construct a joint time-frequency enhanced vital sign map.

[0074] Since environmental noise and random interference are typically independent and uncorrelated in spatial distribution, while the breathing signal of a real target exhibits high spectral consistency across different receiving channels, this embodiment employs multi-channel HPS to enhance spectral cross-correlation for multi-target detection and localization. One channel ( The distance-frequency domain matrix after HPS enhancement can be expressed as: ,in Let the distance cell index be... and They represent The distance unit, The The HPS enhancement spectrum of a distance cell can be expressed as:

[0075] in and The respiratory spectrum corresponds to the two channels. and This represents residual background noise after preprocessing. (In the respiratory-related frequency band) Within the range of 0.1-2Hz, the background noise of the two channels is not only independent of each other but also unrelated to the breathing signal. and cross-correlation coefficient It can be calculated as:

[0076] in To express the covariance, This indicates the calculation of variance.

[0077] To automatically search for the correct pairing of two channels across the entire distance dimension and visually demonstrate the distribution of spectral correlation in the distance dimension, based on the formula... The core idea of ​​this embodiment is to use the cross-correlation coefficients of a single distance gate. Generalize to pairwise correlations of all distance gates, and construct an enhanced spectral correlation matrix. In the respiratory-related frequency band Inside, record , For the two channels in the first HPS enhancement spectrum of a distance gate, These are the mean values ​​within their respective bands. For and After calculating the correlation of all distance gates, a two-dimensional spatiotemporal correlation matrix is ​​generated. :

[0078] in represent and The enhanced spectral correlation coefficients, with the matrix row and column indices corresponding to the distance gates of channel one and channel two, respectively. Equation (11) is a generalization of equation (10). With formula It can be concluded that if a pair of distance gates originate from the same human target, they exhibit strong consistency across the spatial, temporal, and frequency dimensions, and their correlation coefficient will be significantly higher; conversely, the correlation coefficient is lower in background noise regions. Therefore, the matrix... The target pairing location is presented as a clear, bright patch, while the remaining area is a dark background. This contrast allows for correct pairing of the two channels, and this significant contrast enables effective dual-channel target pairing, distance extraction, and localization.

[0079] The correlation matrix obtained based on the above-mentioned spatiotemporal-frequency joint enhancement A two-dimensional constant false alarm rate (CFAR) detection method is used to achieve automatic identification of living targets. Specifically, it employs a two-dimensional rectangular sliding window on the correlation matrix. The process iterates through the area, with an inner guard window and an outer training window centered on the cut pixel. When extracting background pixels, only the pixels within the outer training window are used to estimate the local background mean.

[0080] in Indicates the value taken by the training unit. Represents the total number of training units. Threshold According to the preset false alarm probability calculate:

[0081] If the correlation matrix The currently inspected pixel If the target location is positive, its decision value is set to 1; otherwise, it is set to 0. After CFAR detection, potential target areas can be marked. However, in real-world scenarios, false targets often exist due to residual noise. Furthermore, due to the micro-Doppler effect caused by human breathing, targets typically cover multiple consecutive range cells in the correlation matrix. To effectively eliminate false alarms and accurately locate the target, this embodiment further provides a discrimination method based on energy criteria. Assuming that the range gate with the strongest echo energy is most likely the location of the target, then the... The energy of the distance gate Defined as:

[0082] in, This is the respiratory-related frequency band. No. The HPS-enhanced spectrum corresponding to each range gate. Based on the formula, the energy of each range cell in the range frequency domain matrix after HPS enhancement for both channels is calculated to obtain the corresponding range energy spectrum. Then, the coordinates of all cells in each region are mapped back to these two range energy spectra. The range gate corresponding to the maximum energy value in the region is selected as the location of the candidate target. Subsequently, the range gate is checked in the HPS-enhanced spectrum to see if it contains a valid respiratory peak. If it exists and the frequency deviation of the peaks in the two channels is within the allowable minimum error range, the region is determined to correspond to a real life target, and the range gates corresponding to the region in the two channels are recorded as the distance L from each channel to the target; otherwise, the candidate region is discarded as a false alarm.

[0083] Finally, given the known geometric distribution of the antenna array, the two-dimensional positioning of the target is achieved using an elliptical intersection positioning algorithm. Assume the coordinates of the transmitting antenna are... The coordinates of receiving antenna 1 are The coordinates of receiving antenna 2 are Then the target's coordinates This can be obtained by solving the following system of equations:

[0084] In summary, the process for detecting weak life targets using through-wall radar based on joint space-time-frequency enhancement is as follows:

[0085] (1) Radar echo preprocessing:

[0086] Remove the static background and zero-frequency components from the echo matrix according to equation (5) and perform bandpass filtering.

[0087] (2) Enhancement of time-frequency domain features:

[0088] Perform an FFT on the preprocessed signal along the slow time dimension according to equation (7). Calculate the distance-frequency domain matrix according to equation (8), and use the harmonic product spectrum (HPS) technique to perform multi-scale fusion enhancement of the spectrum.

[0089] (3) Construction of joint spatiotemporal features:

[0090] According to Equation (11), the cross-correlation coefficient between the two channels is calculated using the enhanced spectrum, and a life characteristic correlation matrix with spatiotemporal-frequency joint enhancement is constructed.

[0091] (4) CFAR Automatic Target Detection:

[0092] According to equation (13), set the guard window and training window on the correlation matrix, calculate the local background mean and detection threshold, and mark the potential target area.

[0093] (5) Energy identification and distance pairing:

[0094] The energy spectrum of the candidate distance gate is calculated according to formula (14). Based on the principle of maximizing energy and the consistency of the respiratory peak frequency, false alarms are eliminated and the target distance of the dual channels is extracted.

[0095] (6) Two-dimensional positioning solution:

[0096] Substitute the extracted target distance into equation (15) to solve for the final two-dimensional spatial coordinates of the target.

[0097] To further demonstrate the effectiveness of this embodiment, the following supplementary explanation is based on actual test results.

[0098] 1. Experimental conditions

[0099] The computer used in this embodiment is configured with an Intel Core i7-10875H processor. The software platform is MATLAB 2021b. Radar data was acquired using a MIMO ultra-wideband through-wall radar prototype developed by the research group. This radar adopts a frequency-stepping system, with a 2-transmit, 2-receive antenna array, a frequency range of 800MHz to 1.6GHz, and is placed close to a concrete wall. This embodiment designed two sets of life target detection experiments. In the first set of experiments, a stationary target was located at a relatively far position behind the wall, with coordinates (0m, 8.5m). In the second set of experiments, two stationary targets were located behind the wall, with coordinates (0m, 5m) and (0m, 7.5m) respectively.

[0100] Figure 2 The preprocessing results of the first set of dual-channel signals are presented. It can be observed that when a stationary target is located far behind the wall, the signal attenuation is severe, and living targets are almost impossible to identify. Figure 3 This section presents the time-domain and frequency-domain cross-correlation results, along with the processing results of this embodiment. The time-domain cross-correlation is calculated on the pre-processed echo in the slow time domain, while the frequency-domain cross-correlation is performed directly on the FFT spectrum without HPS enhancement. Correspondingly... Figure 3 (d)-(f) show the results of CFAR detection. It can be observed that when the target is at a considerable distance, neither the time-domain nor the frequency-domain cross-correlation method can accurately detect the target, while this embodiment can correctly detect the target. (Combining the above results with...) The positioning results show that this embodiment can accurately detect targets and achieve precise positioning.

[0101] surface Comparison of localization results and errors (m) of different methods in Experiment 1

[0102] Figure 4 The preprocessing results of the dual-channel signals from the second set of experiments are presented. It can be observed that due to differences in distance and breathing amplitude among the different targets, and because target two is located behind target one and is obstructed by it, the echo energy is extremely weak. Figure 5 It can be found that false targets or missed detections still exist after processing by other methods, while this embodiment detects all human targets. The results of different methods are listed, which show that the invention not only locates all targets but also keeps the positioning error at a low level, which fully demonstrates its excellent multi-target resolution and detection capabilities.

[0103] surface Comparison of positioning results and errors (m) of different methods in Experiment 2

[0104] This embodiment first performs pulse compression and preprocessing on the radar echo in the time domain. Then, coherent accumulation is performed using Fast Fourier Transform (FFT) to transform the signal to the frequency domain. Based on this, multi-scale spectral fusion of the range-Doppler (RD) matrix is ​​achieved using Harmonic Product Spectrum (HPS) technology, effectively enhancing the periodic breathing signal while significantly suppressing random noise. In the spatial domain, this method utilizes the enhanced multi-channel spectrum to construct a cross-channel correlation matrix, generating a vital sign feature map based on spatio-temporal-frequency joint enhancement. Subsequently, by combining constant false alarm rate (CFAR) detection with energy criteria, automatic identification and range matching of stationary targets are achieved. Finally, the two-dimensional (2D) spatial coordinates of the target are calculated using the elliptic cross-location method. This complete "pulse compression—HPS enhancement—cross-correlation—geometric location" processing pipeline solves the problems of false alarms and missed alarms, and poor robustness in existing methods, enabling target detection by through-wall radar.

[0105] This embodiment also provides a life target detection device using stepped-frequency continuous-wave through-wall radar, including:

[0106] The acquisition unit is used to acquire multi-channel fast-time data of step-frequency continuous wave through-wall radar obtained from slow-time dimension observations.

[0107] The first transformation unit is used to perform inverse fast Fourier transform on the fast time data of each channel to obtain a two-dimensional echo matrix. The size of the two-dimensional echo matrix is ​​M×N, where M is the number of fast time sampling points and N is the number of slow time sampling points.

[0108] The second transformation unit is used to perform a fast Fourier transform on each range gate of the two-dimensional echo matrix along the slow time dimension to obtain a range-Doppler matrix;

[0109] The enhancement unit is used to perform harmonic product spectrum processing on the range-Doppler matrix to obtain a range-frequency domain matrix;

[0110] The construction unit is used to construct a multi-channel spatiotemporal correlation matrix based on the distance-frequency domain matrix;

[0111] The traversal unit is used to traverse the multi-channel spatiotemporal-frequency correlation matrix using a two-dimensional constant false alarm rate method, mark potential target regions, and obtain multi-channel target distances; and,

[0112] The solving unit is used to solve for the target coordinates based on the multi-channel target distance.

[0113] This application also provides an electronic device, including: a processor, and a memory coupled to the processor, the memory being used to store a computer program; the processor being used to execute the computer program stored in the memory, so that the electronic device performs the method as described in any of the above embodiments.

[0114] Electronic devices can be computing devices such as desktop computers, laptops, handheld computers, and cloud servers. These electronic devices may include, but are not limited to, processors and memory.

[0115] The processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor. The processor is the control center of the electronic device, connecting various parts of the device via various interfaces and lines.

[0116] The memory can be used to store the computer program, and the processor implements various functions of the electronic device by running or executing the computer program stored in the memory and calling the data stored in the memory.

[0117] The memory may primarily include a program storage area and a data storage area. The program storage area may store the operating system, applications required for at least one function, etc.; the data storage area may store data created based on the use of the mobile phone, etc. In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, memory, plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0118] This application also provides a storage medium, which is a computer-readable storage medium. The computer program is stored in the computer-readable storage medium, and when executed by a processor, the computer program can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable file, or some intermediate form. The computer-readable medium can include: any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium, etc.

[0119] This application also provides a computer program product, including: a computer program or instructions that, when the computer program or instructions are run on a computer, cause the computer to perform any of the above possible implementation methods.

[0120] The above description is the preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications are also considered to be within the scope of protection of this application.

Claims

1. A method for detecting living targets using stepped-frequency continuous-wave through-wall radar, characterized in that, include: Obtain multi-channel fast-time data of step-frequency continuous wave through-wall radar obtained from slow-time dimension observations; Perform an inverse fast Fourier transform on the fast time data of each channel to obtain a two-dimensional echo matrix. The size of the two-dimensional echo matrix is ​​M×N, where M is the number of fast time sampling points and N is the number of slow time sampling points. Perform a Fast Fourier Transform along the slow time dimension on each range gate of the two-dimensional echo matrix to obtain the range-Doppler matrix; The range-Doppler matrix is ​​processed using harmonic product spectrum techniques to obtain a range-frequency domain matrix; Based on the distance-frequency domain matrix, a multi-channel spatiotemporal-frequency correlation matrix is ​​constructed; A two-dimensional constant false alarm rate (CFAR) method is used to traverse the multi-channel spatiotemporal-frequency correlation matrix, mark potential target regions, and obtain multi-channel target distances; and, Based on the multi-channel target distance, the target coordinates are calculated.

2. The method for detecting living targets using stepped-frequency continuous-wave through-wall radar as described in claim 1, characterized in that, The method for detecting living targets using step-frequency continuous wave through-wall radar also includes: Before performing the Fast Fourier Transform, the mean of the slow time series corresponding to each distance gate of the two-dimensional echo matrix is ​​calculated, and the mean is subtracted from the original sequence of the two-dimensional echo matrix to filter out the zero-frequency background component.

3. The method for detecting living targets using stepped-frequency continuous-wave through-wall radar as described in claim 1, characterized in that, The method for detecting living targets using step-frequency continuous wave through-wall radar also includes: A slow-time bandpass filter is applied to the two-dimensional echo matrix, with the bandpass filter range set to 0.1–2 Hz.

4. The method for detecting living targets using stepped-frequency continuous-wave through-wall radar as described in claim 1, characterized in that, The distance-Doppler matrix is ​​expressed as: The distance-frequency domain matrix is ​​expressed as: Where h is the distance gate, f is the frequency, and n is the slow-time sampling point. It is a two-dimensional echo matrix. Representing the distance-Doppler matrix The spectrum of the downsampled version, where m is the distance cell index. This indicates the number of harmonics being considered.

5. The method for detecting living targets using stepped-frequency continuous-wave through-wall radar as described in claim 4, characterized in that, Multi-channel is dual-channel, the first The distance-frequency domain matrix of each channel is expressed as follows: , express The Enhanced spectrum of each distance unit, express The The enhanced spectrum of each distance cell, and the multi-channel spatiotemporal correlation matrix are: Where B represents the respiratory-related frequency band, and f represents the frequency. These are the mean values ​​within their respective bands.

6. The method for detecting living targets using stepped-frequency continuous-wave through-wall radar as described in claim 1, characterized in that, The two-dimensional constant false alarm rate method is used to traverse the multi-channel spatiotemporal-frequency correlation matrix and mark potential target regions, specifically including: A two-dimensional rectangular sliding window is used to traverse the multi-channel spatiotemporal correlation matrix. The window is set with an inner protective window and an outer training window centered on the pixel being inspected. The local background mean is estimated using the pixels in the outer training window. The decision threshold is determined using the local background mean and the preset false alarm probability; and, The potential target region is determined by using the multi-channel spatiotemporal correlation matrix value of the detected pixel and the decision threshold.

7. The method for detecting living targets using stepped-frequency continuous-wave through-wall radar as described in claim 1, characterized in that, The method for detecting life targets using stepped-frequency continuous-wave through-wall radar also includes: After marking potential target regions, the energy value corresponding to each range gate is obtained based on the range-frequency domain matrix; The energy values ​​are used to construct the distance-energy spectrum corresponding to each channel; Map the coordinates of all range cells in each potential target region back to the range energy spectrum of each channel, and select the range gate corresponding to the maximum energy value in the region as the location of the candidate target; and... In the distance-frequency domain matrix, check whether the distance gate corresponding to the maximum energy value contains a valid respiratory peak. If it exists and the frequency deviation of the two channel peaks is within the specified error range, then determine that the potential target area corresponds to a real life target, and record the distance gate corresponding to the potential target area in each channel as the distance L from each channel to the target.

8. The method for detecting living targets using stepped-frequency continuous-wave through-wall radar as described in claim 7, characterized in that, Corresponding to the receiving antenna, based on the multi-channel target distance, the target coordinates (x, y) are calculated as follows: The coordinates of the transmitting antenna are: The coordinates of receiving antenna 1 are The coordinates of receiving antenna 2 are .

9. A life target detection device using stepped-frequency continuous-wave through-wall radar, characterized in that, include: The acquisition unit is used to acquire multi-channel fast-time data of step-frequency continuous wave through-wall radar obtained from slow-time dimension observations. The first transformation unit is used to perform inverse fast Fourier transform on the fast time data of each channel to obtain a two-dimensional echo matrix. The size of the two-dimensional echo matrix is ​​M×N, where M is the number of fast time sampling points and N is the number of slow time sampling points. The second transformation unit is used to perform a fast Fourier transform on each range gate of the two-dimensional echo matrix along the slow time dimension to obtain a range-Doppler matrix; The enhancement unit is used to perform harmonic product spectrum processing on the range-Doppler matrix to obtain a range-frequency domain matrix; The construction unit is used to construct a multi-channel spatiotemporal correlation matrix based on the distance-frequency domain matrix; The traversal unit is used to traverse the multi-channel spatiotemporal-frequency correlation matrix using a two-dimensional constant false alarm rate method, mark potential target regions, and obtain multi-channel target distances; and, The solving unit is used to solve for the target coordinates based on the multi-channel target distance.

10. An electronic device, characterized in that, The electronic device includes: a processor, and a memory coupled to the processor. The memory is used to store computer programs; The processor is configured to execute the computer program stored in the memory, so that the electronic device performs the method for detecting living targets using step-frequency continuous wave through-wall radar as described in any one of claims 1-8.

Images