A device for extracting motion data of different human limbs based on multi-channel radar

Abstract

The present invention belongs to the technical field of radar target motion data detection applications, and specifically relates to a device for extracting motion data of different limbs of a human body based on a multi-channel radar. The device comprises: a multi-channel radar and a data processing terminal arranged on a human target; the multi-channel radar is respectively configured with at least two channels in pitch and azimuth directions, for correspondingly collecting pitch motion data and azimuth motion data of the human target in each channel in real time; the data processing terminal is used to receive the pitch motion data and azimuth motion data collected by each channel in real time, process the data to obtain three-dimensional motion data of speed, distance and time for each channel of the human target; perform coherent superposition and human target detection on the data; perform multi-channel joint processing on the detected three-dimensional motion data of the human target, and separate and extract the motion data of different limbs of the human body by setting a pitch threshold and an azimuth threshold of the human target.

Description

Technical Field

[0001] The present invention belongs to the technical field of motion data detection applications of radar targets and motion data detection technology of wearable devices. Specifically, it relates to a device for extracting motion data of different limbs of the human body based on multi-channel radar. Background Art

[0002] When a human target experiences radial motion relative to the radar, the frequency of the radar echo will produce a Doppler effect. In many cases, in addition to translational motion, any structural component of an object will also experience some oscillatory motion—called micromotion. Examples include the rotating blades of a helicopter, the flapping wings of a bird in flight, and the swinging arms and legs of a pedestrian. Micromotion can cause Doppler frequency modulation of the transmitted radar signal around the carrier frequency. This Doppler frequency modulation is called the micro-Doppler frequency. The micromotion characteristics of different human targets vary, and therefore, the generated micro-Doppler frequencies also vary. Therefore, the unique micro-Doppler frequencies generated by different human targets are called micro-Doppler signatures, which can be used for human target identification and classification. Human targets exhibit typical micromotions during movement. Due to the flexibility of human joints, both the micromotions of human targets and the micro-Doppler signatures they generate are highly complex and unique. Human micro-motions are primarily composed of arm and leg swings. Therefore, a human micro-Doppler signature is a collection of micro-Doppler frequency components caused by different limb movements. Therefore, based on the micro-Doppler signature of the human body, the motion state of the human target can be identified and classified.

[0003] Recently, significant progress has been made in combining human micro-Doppler signatures with classification and recognition algorithms for human motion state classification and gait recognition. However, for micro-motions of different limbs, single-channel radar can only obtain the overall micro-Doppler signature and distance information for one (or more) human targets, but cannot obtain the target's angle information. Therefore, micro-Doppler signatures obtained solely by single-channel radar cannot distinguish and extract the micro-Doppler frequencies contributed by different limbs of the human target. Due to the diversity of human motion states, different motion states may produce similar micro-Doppler signatures (for example, the micro-Doppler signature generated by a swing of the right arm and the micro-Doppler signature generated by a swing of the right arm), leading to confusion in the recognition process based solely on micro-Doppler signatures. Therefore, it is crucial to extract and resolve the different frequency components of the micro-Doppler signature. Multi-channel radar, in addition to obtaining motion data such as the micro-Doppler signature and distance information of a human target, can also obtain the target's angle information, thus enabling it to separate and distinguish the motion data of different human limbs.

[0004] Existing work on multi-channel radar-based human motion detection focuses on identifying, separating, imaging, and tracking multiple people, but has not yet explored extracting the motion data of the different limbs of a single human target. Furthermore, work on the other hand focuses on identifying and classifying the motion data of different limbs as a whole, similarly failing to extract the motion data of different limbs separately. Currently, there is no work on extracting the motion data of different limbs separately using multi-channel radar. Summary of the Invention

[0005] In order to solve the problems existing in the prior art, the present invention proposes a device for extracting motion data of different limbs of the human body based on multi-channel radar. The device can distinguish and separate the motion data of different limbs of the human target, which is not only conducive to higher-precision motion state recognition and classification of the human target, but also can realize three-dimensional reconstruction of the motion state of the human target.

[0006] The present invention provides a device for extracting motion data of different limbs of a human body based on a multi-channel radar, the device comprising: a multi-channel radar and a data processing terminal arranged on a human target;

[0007] The multi-channel radar is configured with at least two channels in the pitch direction and the azimuth direction respectively, for correspondingly collecting the pitch motion data and the azimuth motion data of the human target in each channel in real time, and sending the data to the data processing terminal;

[0008] The data processing terminal is used to receive the pitch motion data and azimuth motion data collected by each channel in real time, process them, and obtain three-dimensional motion data of speed, distance and time for each channel of the human target; coherently superimpose the three-dimensional motion data of each channel and perform human target detection; perform multi-channel joint processing on the detected three-dimensional motion data of the human target to obtain the pitch angle and azimuth angle corresponding to the motion data; and separate and extract the motion data of different human targets by setting the pitch threshold and azimuth threshold of the human target.

[0009] As one of the improvements of the above technical solution, the human target has different limbs; the different limbs include: left arm, right arm, left knee, right knee, left foot and right foot.

[0010] As one of the improvements of the above technical solution, the data processing terminal includes: a three-dimensional motion data acquisition module, a target detection module, a joint processing module and a separation and extraction module;

[0011] The three-dimensional motion data acquisition module is used to receive the pitch motion data and azimuth motion data collected by each channel in real time, process them, and obtain the speed, distance and time three-dimensional motion data of each channel of the human target;

[0012] The target detection module is used to coherently superimpose the three-dimensional motion data of each channel and perform human target detection;

[0013] The joint processing module is used to perform multi-channel joint processing on the detected three-dimensional motion data of the human target to obtain the pitch angle and azimuth angle corresponding to the motion data;

[0014] The separation and extraction module is used to separate and extract motion data of different human targets by setting the pitch threshold and the azimuth threshold of the human targets.

[0015] As one of the improvements of the above technical solution, the three-dimensional motion data acquisition module includes:

[0016] A data receiving unit, configured to receive in real time the pitch motion data and azimuth motion data collected by each channel; and

[0017] The three-dimensional motion data processing unit is used to perform range pulse compression processing on the pitch motion data and azimuth motion data collected by each channel when the number of channels of the multi-channel radar is N, obtain the distance-time information of the human target in each channel, and then perform time-frequency analysis processing in the slow time domain to obtain the speed-time information of the human target in each channel; integrate the distance-time information of the human target in each channel and the speed-time information of the human target in each channel, and finally obtain the three-dimensional motion data S of the distance, speed and time of the human target in each channel. i (r, v, t); where the subscript i represents the i-th channel; i = 1, 2, …, N.

[0018] As one of the improvements to the above technical solution, the specific implementation process of the target detection module includes:

[0019] Assume that the three-dimensional motion data of the distance, speed and time of the human target in each channel is S i (r,v,t);

[0020] Perform coherent superposition processing on the three-dimensional motion data of the human target's distance r, velocity v and time t in each channel to obtain the superposition result S(r,v,t):

[0021]

[0022] Then, target detection processing is performed on S(r,v,t), and the three-dimensional motion data that does not meet the target detection conditions is eliminated, and the three-dimensional motion data S of the human target that meets the target detection conditions is detected. e (r,v,t):

[0023]

[0024] Among them, r m is the mth element of the distance dimension r vector in the three-dimensional motion data; v n is the nth element of the velocity dimension v vector in the three-dimensional motion data; t p is the pth element of the time dimension t vector in the three-dimensional motion data;

[0025] The target detection condition is to judge the elements S(r,v,t) in the result S(r,v,t) after superposition processing in sequence. m ,v n ,t p ) is greater than the preset amplitude threshold th s , take m=1,2,3,…,M, n=1,2,3,…,N, p=1,2,3,…,P, and judge whether

[0026] abs[S(r m ,v n ,t p )]>th s

[0027] If abs[S(r m ,v n ,t p )]≤th s , it is determined that the target detection conditions are not met and it is eliminated;

[0028] If abs[S(r m ,v n ,t p )]>th s , then it is determined that the target detection condition is met and the element S(r m ,v n ,t p ); and store the data elements S(r,v,t) that meet the target detection conditions in the result S(r,v,t) after superposition processing m ,v n ,t p ) in the subscript Index e =[r e ,v e ,t e ], used to extract the three-dimensional motion data S of each channel i (r, v, t) 3D motion data S after target detection ei (r,v,t) where r e =[r me ,r me+1 ,r me+2 ,…,r Me ],me≥1,andMe≤M;ve =[v ne ,v ne+1 ,v ne+2 ,…,v Ne ], ne≥1, and Ne≤N, t e =t,

[0029] For each channel of three-dimensional motion data S i (r,v,t) according to the stored data element index Index e , extract the three-dimensional motion data S of i channels after target detection in sequence ei (r,v,t):

[0030]

[0031] As one of the improvements of the above technical solution, the joint processing module includes: an azimuth position acquisition unit and a pitch position acquisition unit;

[0032] The position acquisition unit is used to obtain the three-dimensional motion data S of each channel after target detection. ei (r, v, t), which includes the azimuth three-dimensional motion data S eip (r, v, t) and pitch three-dimensional motion data S eiq (r,v,t);

[0033] Select azimuth three-dimensional motion data S eip 3D motion data S of channel m in (r,v,t) eipm (r, v, t) and three-dimensional motion data S of channel n eipn (r, v, t) is conjugate multiplied to obtain the phase angle, and the phase difference information Δφ1 between the two channel signals is:

[0034] Δφ1=angle{S eipm (r,v,t)×conj[S ejpn (r,v,t)]}

[0035] Based on the obtained phase difference Δφ1, the azimuth angle θ1 of the human target relative to the radar is calculated:

[0036]

[0037] Where d is the baseline length between channels, and λ is the wavelength of the radar transmission signal;

[0038] Then we can get the position x1 of the human target relative to the radar:

[0039]

[0040] Where R is the distance from the human target to the radar;

[0041] The pitch position acquisition unit is used to select the pitch direction three-dimensional motion data S eiq 3D motion data S of channel m in (r,v,t) eiqm (r, v, t) and three-dimensional motion data S of channel n eiqn (r, v, t) is conjugate multiplied to obtain the phase angle, and the phase difference information Δφ2 between the two channel signals is:

[0042] Δφ2=angle{S eiqm (r,v,t)×conj[S ejqn (r,v,t)]}

[0043] Based on the obtained phase difference Δφ2, the azimuth angle θ2 of the human target relative to the radar is calculated:

[0044]

[0045] Where d is the baseline length between channels, and λ is the wavelength of the radar transmission signal;

[0046] Then we can get the elevation position x2 of the human target relative to the radar:

[0047]

[0048] Where R is the distance from the human target to the radar.

[0049] As one of the improvements of the above technical solution, the separation and extraction module specifically includes: a threshold setting unit, a data receiving unit and a judgment unit;

[0050] The threshold setting unit is used to set the azimuth threshold th of the human target relative to the radar. x , used to distinguish the left and right limbs of human targets; set the pitch threshold th of the human target relative to the radar y1 , used to distinguish the upper and lower body of human targets; set the pitch threshold th of the human target relative to the radar y2 , where th y1 >th y2 , used to distinguish the knees and feet of the lower body of a human target;

[0051] The data receiving unit is used to receive the azimuth position x1 of the human target relative to the radar and the pitch position x2 of the human target relative to the radar corresponding to each three-dimensional motion data, and send them to the judgment unit;

[0052] The judgment unit is used to judge the azimuth position x1 of the human target relative to the radar and the pitch position x2 of the human target relative to the radar corresponding to each received three-dimensional motion data;

[0053] If the position x1 of the human target corresponding to each three-dimensional motion data relative to the radar is less than or equal to th x , that is, x1≤th x , and the pitch position x2 is greater than or equal to th y1 , that is, x2≥th y1 , then the three-dimensional motion data belongs to the left arm of the human target;

[0054] If the position x1 of the human target corresponding to each 3D motion data obtained relative to the radar is greater than th x , that is, x1>th x , and the pitch position x2 is greater than or equal to th y1 , that is, x2≥th y1 , then the three-dimensional motion data belongs to the right arm of the human target;

[0055] If the position x1 of the human target corresponding to each three-dimensional motion data relative to the radar is less than or equal to th x , that is, x1≤th x , and the pitch position x2 is less than th y1 At the same time greater than th y2 , that is, th y2 <x2<th y1 , then the three-dimensional motion data belongs to the left knee of the human target;

[0056] If the position x1 of the human target corresponding to each 3D motion data obtained relative to the radar is greater than th x , that is, x1>th x , and the pitch position x2 is less than th y1 At the same time greater than th y2 , that is, th y2 <x2<th y1 , then the three-dimensional motion data belongs to the right knee of the human target;

[0057] If the position x1 of the human target corresponding to each three-dimensional motion data relative to the radar is less than or equal to th x , that is, x1≤th x , and the pitch position x2 is less than or equal to th y2 , that is, x2≤th y2 , then the three-dimensional motion data belongs to the left foot of the human target;

[0058] If the position x1 of the human target corresponding to each 3D motion data obtained relative to the radar is greater than th x , that is, x1>th x, and the pitch position x2 is less than or equal to th y2 , that is, x2≤th y2 , then the three-dimensional motion data belongs to the right foot of the human target;

[0059] By performing the above-mentioned threshold judgment on the 3D motion data based on the azimuth position and pitch position corresponding to the 3D motion data of each channel, the 3D motion data can be separated and extracted from the 3D motion data of different limbs of the human target.

[0060] The beneficial effects of the present invention compared with the prior art are:

[0061] The device of the present invention can distinguish and separate the motion data of different limbs of a human target by acquiring angle information through a multi-channel radar. This not only solves the confusion problem in target recognition and classification caused by similar micro-Doppler characteristics generated by different motion states of a human target, and improves the accuracy of recognition and classification of the motion state of a human target, but also can be used to quantitatively analyze the motion data of each limb of a human target and realize three-dimensional reconstruction of the motion of different limbs of a human target. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 It is a principle diagram of the working process of the device for extracting motion data of different limbs of the human body based on multi-channel radar of the present invention;

[0063] Figure 2 2. It is a schematic structural diagram of a device for extracting motion data of different limbs of a human body based on a multi-channel radar for collecting radar human motion data of the present invention;

[0064] Figure 3 This is the result of the radar echo data of the human target in the state of swinging arms in place after the range pulse compression;

[0065] Figure 4 This is the result diagram of moving target detection after the radar echo data in the stationary swing arm state is compressed in the range direction;

[0066] Figure 5 The micro-Doppler characteristic diagram is obtained by performing time-frequency analysis on the radar echo data in the stationary swing arm state in the slow time domain;

[0067] Figure 6 It is a three-dimensional motion data display diagram of the distance, speed and time of the radar echo data in the swing arm state after CFAR target detection;

[0068] Figure 7 It is a four-dimensional data display diagram that combines the azimuth position information obtained through azimuth multi-channel processing with the three-dimensional motion data;

[0069] Figure 8It is a four-dimensional data display diagram that combines the pitch position information obtained through multi-channel pitch processing with the three-dimensional motion data;

[0070] Figure 9 This is a schematic diagram of the ideal threshold setting;

[0071] Figure 10 This is a display of the three-dimensional motion data of the left arm separated after threshold judgment of the three-dimensional motion data;

[0072] Figure 11 This is a display of the three-dimensional motion data of the right arm separated after threshold judgment of the three-dimensional motion data;

[0073] Figure 12 This is a three-dimensional motion data display diagram of the distance, speed and time of the stationary step data after CFAR target detection;

[0074] Figure 13 It is a display diagram of the three-dimensional motion data of the arm that is separated and extracted after threshold judgment of the three-dimensional motion data;

[0075] Figure 14 This is a display of the three-dimensional motion data of the knee that is separated and extracted after threshold judgment of the three-dimensional motion data;

[0076] Figure 15 This is a display diagram of the three-dimensional motion data of the foot that is separated and extracted after threshold judgment of the three-dimensional motion data. DETAILED DESCRIPTION

[0077] The present invention will now be further described with reference to the accompanying drawings.

[0078] The present invention provides a device for extracting motion data of different limbs of the human body based on a multi-channel radar, which solves the problem that the motion information of different limbs of the human body is difficult to distinguish and separate when the radar obtains human motion data.

[0079] like Figure 1 and 2 As shown, the device includes: a multi-channel radar set in the pitch direction of the human target, a multi-channel radar set in the azimuth direction of the human target and a data processing terminal;

[0080] A multi-channel radar is set in the azimuth direction of the human target to collect the azimuth movement data of the human target in real time and send it to the data processing terminal;

[0081] The data processing terminal is used to receive the pitch motion data and azimuth motion data collected by each channel in real time, process them, and obtain the three-dimensional motion data of speed, distance and time of each channel about the human target; coherently superimpose the three-dimensional motion data of each channel to obtain the result of coherent superposition of each channel, and perform human target detection on it; perform multi-channel joint processing on the detected three-dimensional motion data of the human target to obtain the pitch angle (pitch position) and azimuth angle (azimuth position) corresponding to the motion data; and distinguish and separate the motion data of different human limbs by setting pitch thresholds and azimuth thresholds.

[0082] In this embodiment, the human target is a human target, and the human target movement states include: walking, running, jumping, falling, and waving; the different limbs of the human body include: left arm, right arm, left knee, right knee, left foot, and right foot.

[0083] Specifically, the data processing terminal includes: a three-dimensional motion data acquisition module, a target detection module, a joint processing module and a separation and extraction module;

[0084] The three-dimensional motion data acquisition module is used to receive the pitch motion data and azimuth motion data collected by each channel in real time, process them, and obtain the speed, distance and time three-dimensional motion data of each channel of the human target;

[0085] Specifically, the three-dimensional motion data acquisition module includes:

[0086] A data receiving unit, configured to receive in real time the pitch motion data and azimuth motion data collected by each channel; and

[0087] The three-dimensional motion data processing unit is used to perform range pulse compression processing on the pitch motion data and azimuth motion data collected by each channel when the number of channels of the multi-channel radar is N, obtain the distance-time information of the human target in each channel, and then perform time-frequency analysis processing in the slow time domain to obtain the speed-time information of the human target in each channel; integrate the distance-time information of the human target in each channel and the speed-time information of the human target in each channel, and finally obtain the three-dimensional motion data S of the distance, speed and time of the human target in each channel. i (r, v, t); where the subscript i represents the i-th channel; i = 1, 2, …, N.

[0088] The target detection module is used to coherently superimpose the three-dimensional motion data of each channel and perform human target detection;

[0089] Specifically, the specific implementation process of the target detection module includes:

[0090] Assume that the three-dimensional motion data of the distance, speed and time of the human target in each channel is S i (r,v,t);

[0091] Perform coherent superposition processing on the three-dimensional motion data of the human target's distance r, velocity v and time t in each channel to obtain the superposition result S(r,v,t):

[0092]

[0093] Then, target detection processing is performed on S(r,v,t), such as CFAR (Constant False Alarm Rate) constant false alarm detection, to eliminate the three-dimensional motion data that does not meet the target detection conditions, and detect the three-dimensional motion data S of the human target that meets the target detection conditions. e (r,v,t):

[0094]

[0095] Among them, r m is the mth element of the distance dimension r vector in the three-dimensional motion data (r=[r1,r2,r3,…,r m ,r m+1 ,…,r M ]);v n is the nth element of the velocity dimension v vector in the three-dimensional motion data (v=[v1,v2,v3,…,v n ,v n+1 ,…,v N ]); t p is the pth element of the time dimension t vector in the three-dimensional motion data (t=[t1,t2,t3,…,t p ,t p+1 ,…,t P ]);

[0096] The target detection condition is to judge the elements S(r,v,t) in the result S(r,v,t) after superposition processing in sequence. m ,v n ,t p ) is greater than the preset amplitude threshold th s , take m=1,2,3,…,M, n=1,2,3,…,N, p=1,2,3,…,P, and judge whether

[0097] abs[S(r m ,v n ,t p )]>th s

[0098] If abs[S(rm ,v n ,t p )]≤th s , it is determined that the target detection conditions are not met and it is eliminated;

[0099] If abs[S(r m ,v n ,t p )]>th s , then it is determined that the target detection condition is met and the element S(r m ,v n ,t p ); and store the data elements S(r,v,t) that meet the target detection conditions in the result S(r,v,t) after superposition processing m ,v n ,y p ) in the subscript Index e =[r e ,v e ,t e ], used to extract the three-dimensional motion data S of each channel i (r, v, t) 3D motion data S after target detection ei (r,v,t) where r e =[r me ,r me+1 ,r me+2 ,…,r Me ],me≥1,andMe≤M;v e =[v ne ,v ne+1 ,v ne+2 ,…,v Ne ], ne≥1, and Ne≤N, t e =t,

[0100] For each channel of three-dimensional motion data S i (r,v,t) according to the stored data element index Index e , extract the three-dimensional motion data S of i channels after target detection in sequence ei (r, v, t) for multiple channel signal processing to extract azimuth and elevation position information:

[0101]

[0102] The joint processing module is used to perform multi-channel joint processing on the detected three-dimensional motion data of the human target to obtain the pitch angle and azimuth angle corresponding to the motion data;

[0103] Specifically, the joint processing module specifically includes: an azimuth position acquisition unit and a pitch position acquisition unit;

[0104] The multi-channel data in azimuth and elevation are processed separately to obtain azimuth angle (azimuth position) information and elevation angle (elevation position) information.

[0105] Specifically, the position acquisition unit is used to obtain the three-dimensional motion data S of each channel after target detection. ei (r, v, t), which includes the azimuth three-dimensional motion data S eip (r, v, t) and pitch three-dimensional motion data S eiq (r,v,t);

[0106] Select azimuth three-dimensional motion data S eip 3D motion data S of channel m in (r,v,t) eipm (r, v, t) and three-dimensional motion data S of channel n eipn (r, v, t) is conjugate multiplied to obtain the phase angle, and the phase difference information Δφ1 between the two channel signals is:

[0107] Δφ1=angle{S eim (r,v,t)×conj[S ejpn (r,v,t)]}

[0108] Based on the obtained phase difference Δφ1, the azimuth angle θ1 of the human target relative to the radar is calculated:

[0109]

[0110] Where d is the baseline length between channels, and λ is the wavelength of the radar transmission signal;

[0111] Then we can get the position x1 of the human target relative to the radar:

[0112]

[0113] Where R is the distance from the human target to the radar;

[0114] The pitch position acquisition unit is used to select the pitch direction three-dimensional motion data S eiq 3D motion data S of channel m in (r,v,t) eiqm (r, v, t) and three-dimensional motion data S of channel n eiqn (r, v, t) is conjugate multiplied to obtain the phase angle, and the phase difference information Δφ2 between the two channel signals is:

[0115] Δφ2=angle{S eiqm (r,v,t)×conj[S ejqn (r,v,t)]}

[0116] Based on the obtained phase difference Δφ2, the pitch angle θ2 of the human target relative to the radar is calculated:

[0117]

[0118] Where d is the baseline length between channels, and λ is the wavelength of the radar transmission signal;

[0119] Then we can get the elevation position x2 of the human target relative to the radar:

[0120]

[0121] Where R is the distance from the human target to the radar.

[0122] The separation and extraction module is used to separate and extract motion data of different human targets by setting the pitch threshold and the azimuth threshold of the human targets.

[0123] Specifically, the separation and extraction module includes: a threshold setting unit, a data receiving unit and a judgment unit;

[0124] The threshold setting unit is used to set the azimuth threshold th of the human target relative to the radar. x , used to distinguish the left and right limbs of human targets; set the pitch threshold th of the human target relative to the radar y1 , used to distinguish the upper and lower body of human targets; set the pitch threshold th of the human target relative to the radar y2 , where th y1 >th y2 , used to distinguish the knees and feet of the lower body of a human target;

[0125] The data receiving unit is used to receive the azimuth position x1 of the human target relative to the radar and the pitch position x2 of the human target relative to the radar corresponding to each three-dimensional motion data, and send them to the judgment unit;

[0126] The judging unit is used to judge the azimuth position x1 of the human target relative to the radar and the elevation position x2 of the human target relative to the radar corresponding to each obtained three-dimensional motion data;

[0127] If the position x1 of the human target corresponding to each three-dimensional motion data relative to the radar is less than or equal to th x , that is, x1≤th x , and the pitch position x2 is greater than or equal to th y1 , that is, x2≥th y1 , then the three-dimensional motion data belongs to the left arm of the human target;

[0128] If the position x1 of the human target corresponding to each 3D motion data obtained relative to the radar is greater than th x , that is, x1>th x , and the pitch position x2 is greater than or equal to th y1 , that is, x2≥th y1 , then the three-dimensional motion data belongs to the right arm of the human target;

[0129] If the position x1 of the human target corresponding to each three-dimensional motion data relative to the radar is less than or equal to th x , that is, x1≤th x , and the pitch position x2 is less than th y1 At the same time greater than th y2 , that is, th y2 <x2<th y1 , then the three-dimensional motion data belongs to the left knee of the human target;

[0130] If the position x1 of the human target corresponding to each 3D motion data obtained relative to the radar is greater than th x , that is, x1>th x , and the pitch position x2 is less than th y1 At the same time greater than th y2 , that is, th y2 <x2<th y1 , then the three-dimensional motion data belongs to the right knee of the human target;

[0131] If the position x1 of the human target corresponding to each three-dimensional motion data relative to the radar is less than or equal to th x , that is, x1≤th x , and the pitch position x2 is less than or equal to th y2 , that is, x2≤th y2 , then the three-dimensional motion data belongs to the left foot of the human target;

[0132] If the position x1 of the human target corresponding to each 3D motion data obtained relative to the radar is greater than th x , that is, x1>th x , and the pitch position x2 is less than or equal to th y2 , that is, x2≤th y2 , then the three-dimensional motion data belongs to the right foot of the human target;

[0133] By performing the above-mentioned threshold judgment on the 3D motion data based on the azimuth position and pitch position corresponding to the 3D motion data of each channel, the 3D motion data can be separated and extracted from the 3D motion data of different limbs of the human target.

[0134] Multi-channel radars have angular resolution capabilities. When two or more human targets are at the same distance from the radar, a single-channel radar cannot distinguish and separate them. However, multi-channel radars can combine angular information to distinguish and separate two or more human targets at different angles, thus providing enhanced target detection capabilities. Existing work using multi-channel radars to detect different limb movements of human targets has demonstrated that multi-channel radars have higher recognition and classification accuracy than single-channel radars and can achieve separation, imaging, and tracking of multiple human targets. Therefore, human motion detection based on multi-channel radars has enormous application potential.

[0135] The present invention aims to distinguish and separate the motion data components of individual limbs of a human subject. It proposes a device for extracting motion data from different limbs of a human subject based on multi-channel radar. By distinguishing and separating the three-dimensional motion data of each limb of a human subject, the problem of similar micro-Doppler signatures generated by different motion states, which can cause confusion in motion state identification based on micro-Doppler signatures, is resolved. Consequently, based on the distinguished and separated three-dimensional motion data of each limb of a human subject, higher-precision and robust motion state identification and classification can be achieved. Furthermore, quantitative analysis of the three-dimensional motion data of each limb of a human subject and three-dimensional reconstruction of human motion can be achieved. The device combines the three-dimensional motion data of distance, velocity, and time acquired by the multi-channel radar with the angular information acquired by multiple channels to distinguish and separate the three-dimensional motion data of each limb of a human subject. Experimental data validates the effectiveness of the proposed method.

[0136] Figure 2 Schematic diagram of the experimental scenario for radar human motion data collection. A multi-channel radar is placed at position P1 (0, 0, 1m), collecting elevation and azimuth motion data of a human target 1m above the ground. The target stands within the radar's beam, with both feet at position P2 (0, y(t), 0) on the ground. In this experiment, y(t) = 3m.

[0137] Figure 3 This is the result of range-direction pulse compression of motion data collected by a multi-channel radar of a human target swinging its arms in place. It can be seen that due to the influence of stationary clutter in the environment, the data of the human target swinging its arms in place is difficult to observe. Therefore, human target detection processing is required to eliminate the stationary background clutter and better obtain the human target's motion data.

[0138] Figure 4The figure shows the result of the radar echo data in the stationary swing state after the range pulse compression and the moving target detection processing. After the moving target detection, the distance-time information of the human target in the stationary swing state can be observed. For the convenience of display, the Figure 4 The data in the dotted box are further processed.

[0139] Figure 5 The micro-Doppler signature is obtained by performing slow-time domain time-frequency analysis on the arm-swinging data. This provides the velocity-time information of the radar echo data when the human target is swinging its arms in place. In the micro-Doppler signature, the velocity of both arms is zero at the same moment (when the arms swing to their highest point), resulting in intersections that make it difficult to distinguish and separate the motion data of the two arms.

[0140] Figure 6 This diagram shows the three-dimensional motion data of distance, velocity, and time for radar echo data from a stationary swinging arm after CFAR target detection. Micro-Doppler features are used to add distance information, providing three-dimensional motion data of distance, velocity, and time. Because the distance between the two arms is significantly different when their velocity is zero, the motion data for the two arms no longer intersect. Therefore, this three-dimensional motion data facilitates the resolution and separation of the motion data for each arm.

[0141] Figure 7 This is a 4D data display of the azimuth position information obtained through azimuth multi-channel processing combined with 3D motion data. The grayscale value of the point cloud represents the azimuth position information relative to the radar.

[0142] Figure 8 This is a 4D data display combining pitch position information obtained through multi-channel pitch processing with 3D motion data. The grayscale values ​​in the point cloud represent the pitch position relative to the radar. Furthermore, the resulting azimuth and pitch position information is used to distinguish and separate the 3D motion data of different human limbs.

[0143] Figure 9 The diagram below shows the ideal threshold setting. By applying thresholds to the orientation and pitch information of the 3D motion data, the 3D motion data of different limbs of the human body can be distinguished and separated.

[0144] Figure 10 This is a diagram showing the three-dimensional motion data of the left arm separated after threshold judgment of the three-dimensional motion data.

[0145] Figure 11This figure shows the 3D motion data of the right arm separated after thresholding the 3D motion data. It can be seen that after thresholding the 3D motion data, the 3D motion data of different limbs of the human body during the stationary arm swinging motion was successfully separated.

[0146] Figure 12 This is a three-dimensional motion data display diagram of the distance, speed and time of the stationary step data after CFAR target detection. Figure 6 The data given is the arm swing data in place. In the movement of marching in place, not only the arms but also the legs swing.

[0147] Figure 13 This is a diagram showing the three-dimensional motion data of the arm that is separated and extracted after threshold judgment of the three-dimensional motion data.

[0148] Figure 14 This is a diagram showing the three-dimensional motion data of the knee that is separated and extracted after threshold judgment of the three-dimensional motion data.

[0149] Figure 15 The following figure shows the 3D motion data of the foot after thresholding the 3D motion data. It can be seen that the 3D motion data of different limbs of the human body during the stationary stepping movement were successfully separated after thresholding the 3D motion data.

[0150] The method proposed in the present invention for distinguishing and separating the motion data of different limbs of a human target based on multi-channel radar is applicable to micro-Doppler features (speed-time information) obtained using time-frequency analysis methods such as short-time Fourier transform, Gabor transform, and WVD (Wigner-Ville Distribution). Although the CFAR target detection method is used in target detection, it is also applicable to extended methods based on the CFAR target detection method and other target detection methods. In addition, the method proposed in the present invention is not limited to two or more channels. At least two channels are required for azimuth and pitch respectively, that is, at least three L-shaped channels (azimuth and pitch share one channel) are required to achieve simultaneous azimuth and pitch angle estimation. The more channels, the better the performance. At the same time, although the motion state of the human target analyzed by the present invention is swinging arms in place and stepping in place, the present invention is also applicable to other human motion states, such as walking, running, falling, waving, etc. Furthermore, while the present invention only uses a single human subject for analysis, it is equally applicable to situations where multiple human subjects are simultaneously in different or identical motion states, such as two people walking toward each other or three people walking, running, and waving. The proposed method has important practical implications for extracting and quantitatively analyzing the motion states of individual limbs or joints of a human subject, facilitating higher-precision and robust recognition of different limb motions of a human subject, and enabling three-dimensional reconstruction of the motions of different limbs of the human subject.

[0151] Finally, it should be noted that the above embodiments are intended only to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the embodiments, it should be understood by those skilled in the art that modifications or equivalent substitutions to the technical solutions of the present invention do not depart from the spirit and scope of the technical solutions of the present invention and are intended to be encompassed by the claims of the present invention.

Claims

1. A device for extracting motion data of different limbs of the human body based on multi-channel radar, characterized in that: The device comprises: a multi-channel radar and a data processing terminal arranged on a human target; The multi-channel radar is configured with at least two channels in the pitch direction and the azimuth direction respectively, for correspondingly collecting the pitch motion data and the azimuth motion data of the human target in each channel in real time, and sending the data to the data processing terminal; The data processing terminal is used to receive the pitch motion data and azimuth motion data collected by each channel in real time, process them, and obtain three-dimensional motion data of speed, distance and time for each channel of the human target; coherently superimpose the three-dimensional motion data of each channel and perform human target detection; perform multi-channel joint processing on the detected three-dimensional motion data of the human target to obtain the pitch angle and azimuth angle corresponding to the motion data; and separate and extract the motion data of different human targets by setting the pitch threshold and azimuth threshold of the human target.

2. The device for extracting motion data of different limbs of a human body based on multi-channel radar according to claim 1, characterized in that: The human target has different limbs; the different limbs include: left arm, right arm, left knee, right knee, left foot and right foot.

3. The device for extracting motion data of different limbs of a human body based on multi-channel radar according to claim 1, characterized in that: The data processing terminal includes: a three-dimensional motion data acquisition module, a target detection module, a joint processing module and a separation and extraction module; The three-dimensional motion data acquisition module is used to receive the pitch motion data and azimuth motion data collected by each channel in real time, process them, and obtain the speed, distance and time three-dimensional motion data of each channel of the human target; The target detection module is used to coherently superimpose the three-dimensional motion data of each channel and perform human target detection; The joint processing module is used to perform multi-channel joint processing on the detected three-dimensional motion data of the human target to obtain the pitch angle and azimuth angle corresponding to the motion data; The separation and extraction module is used to separate and extract motion data of different human targets by setting the pitch threshold and the azimuth threshold of the human targets.

4. The device for extracting motion data of different limbs of a human body based on multi-channel radar according to claim 3, characterized in that: The three-dimensional motion data acquisition module includes: A data receiving unit, configured to receive in real time the pitch motion data and azimuth motion data collected by each channel; and The three-dimensional motion data processing unit is used to perform range pulse compression processing on the pitch motion data and azimuth motion data collected by each channel when the number of channels of the multi-channel radar is N, obtain the distance-time information of the human target in each channel, and then perform time-frequency analysis processing in the slow time domain to obtain the speed-time information of the human target in each channel; integrate the distance-time information of the human target in each channel and the speed-time information of the human target in each channel, and finally obtain the three-dimensional motion data S of the distance r, speed v and time t of the human target in each channel. i (r, v, t); where the subscript i represents the i-th channel; i = 1, 2, …, N.

5. The device for extracting motion data of different limbs of a human body based on multi-channel radar according to claim 3, characterized in that: The specific implementation process of the target detection module includes: Assume that the three-dimensional motion data of the distance, speed and time of the human target in each channel is S i (r,v,t); Perform coherent superposition processing on the three-dimensional motion data of the human target's distance r, velocity v and time t in each channel to obtain the superposition result S(r,v,t): Then, target detection processing is performed on S(r,v,t), and the three-dimensional motion data that does not meet the target detection conditions is eliminated, and the three-dimensional motion data S of the human target that meets the target detection conditions is detected. e (r,v,t): Among them, r m is the mth element of the distance dimension r vector in the three-dimensional motion data; v n is the nth element of the velocity dimension v vector in the three-dimensional motion data; t p is the pth element of the time dimension t vector in the three-dimensional motion data; The target detection condition is to judge the elements S(r,v,t) in the result S(r,v,t) after superposition processing in sequence. m ,v n ,t p ) is greater than the preset amplitude threshold th s , take m=1,2,3,…,M, n=1,2,3,…,N, p=1,2,3,…,P, and judge whether abs[S(r m ,v n ,t p )]>th s If abs[S(r m ,v n ,t p )]≤th s , it is determined that the target detection conditions are not met and it is eliminated; If abs[S(r m ,v n ,t p )]>th s , then it is determined that the target detection condition is met and the element S(r m ,v n ,t p ); and store the data elements S(r,v,t) that meet the target detection conditions in the result S(r,v,t) after superposition processing m ,v n ,t p ) in the subscript Index e =[r e ,v e ,t e ], used to extract the three-dimensional motion data S of each channel i (r, v, t) 3D motion data S after target detection ei (r,v,t) where r e =[r me ,r me+1 ,r me+2 ,…,r Me ],me≥1,andMe≤M;v e =[v ne ,v ne+1 ,v ne+2 ,…,v Ne ], ne≥1, and Ne≤N, t e =t, For each channel of three-dimensional motion data S i (r,v,t) according to the stored data element index Index e , extract the three-dimensional motion data S of i channels after target detection in sequence ei (r,v,t):

6. The device for extracting motion data of different limbs of a human body based on multi-channel radar according to claim 3, characterized in that: The joint processing module includes: an azimuth position acquisition unit and a pitch position acquisition unit; The position acquisition unit is used to obtain the three-dimensional motion data S of each channel after target detection. ei (r, v, t), which includes the azimuth three-dimensional motion data S eip (r, v, t) and pitch three-dimensional motion data S eiq (r,v,t); Select azimuth three-dimensional motion data S eip 3D motion data S of channel m in (r, v, t) eipm (r, v, t) and three-dimensional motion data S of channel n eipn (r, v, t) is conjugate multiplied to obtain the phase angle, and the phase difference information Δφ1 between the two channel signals is: Δφ1=angle{S eipm (r,v,t)×conj[S ejpn (r,v,t)]} Based on the obtained phase difference Δφ1, the azimuth angle θ1 of the human target relative to the radar is calculated: Where d is the baseline length between channels, and λ is the wavelength of the radar transmission signal; Then we can get the position x1 of the human target relative to the radar: Where R is the distance from the human target to the radar; The pitch position acquisition unit is used to select the pitch direction three-dimensional motion data S eiq 3D motion data S of channel m in (r, v, t) eiqm (r, v, t) and three-dimensional motion data S of channel n eiqn (r, v, t) is conjugate multiplied to obtain the phase angle, and the phase difference information Δφ2 between the two channel signals is: Δφ2=angle{S eiqm (r,v,t)×conj[S ejqn (r,v,t)]} Based on the obtained phase difference Δφ2, the pitch angle θ2 of the human target relative to the radar is calculated: Where d is the baseline length between channels, and λ is the wavelength of the radar transmission signal; Then we can get the elevation position x2 of the human target relative to the radar: Where R is the distance from the human target to the radar.

7. The device for extracting motion data of different human limbs based on multi-channel radar according to claim 3, characterized in that: The separation and extraction module specifically includes: a threshold setting unit, a data receiving unit and a judgment unit; The threshold setting unit is used to set the position threshold th of the human target relative to the radar. x , used to distinguish the left and right limbs of human targets; set the pitch threshold th of the human target relative to the radar y1 , used to distinguish the upper and lower body of human targets; set the pitch threshold th of the human target relative to the radar y2 , where th y1 >th y2 , used to distinguish the knees and feet of the lower body of a human target; The data receiving unit is used to receive the azimuth position x1 of the human target relative to the radar and the pitch position x2 of the human target relative to the radar corresponding to each three-dimensional motion data, and send them to the judgment unit; The judgment unit is used to judge the azimuth position x1 of the human target relative to the radar and the pitch position x2 of the human target relative to the radar corresponding to each received three-dimensional motion data; If the position x1 of the human target corresponding to each 3D motion data obtained is less than or equal to the radar position x1 x , and the pitch position x2≥th y1 , then the three-dimensional motion data belongs to the left arm of the human target; If the position x1 of the human target corresponding to each 3D motion data relative to the radar is greater than x , and the pitch position x2≥th y1 , then the three-dimensional motion data belongs to the right arm of the human target; If the position x1 of the human target corresponding to each 3D motion data obtained is less than or equal to the radar position x1 x , and the pitch position th y2 <x2<th y1 , then the three-dimensional motion data belongs to the left knee of the human target; If the position x1 of the human target corresponding to each 3D motion data relative to the radar is greater than x , and the pitch position th y2 <x2<th y1 , then the three-dimensional motion data belongs to the right knee of the human target; If the position x1 of the human target corresponding to each 3D motion data obtained is less than or equal to the radar position x1 x , and the pitch position x2≤th y2 , then the three-dimensional motion data belongs to the left foot of the human target; If the position x1 of the human target corresponding to each 3D motion data relative to the radar is greater than x , and the pitch position x2≤th y2 , then the three-dimensional motion data belongs to the right foot of the human target; By performing the above-mentioned threshold judgment on the 3D motion data based on the azimuth position and pitch position corresponding to the 3D motion data of each channel, the 3D motion data can be separated and extracted from the 3D motion data of different limbs of the human target.

Images