CN107024685A - A kind of gesture identification method based on apart from velocity characteristic - Google Patents

A kind of gesture identification method based on apart from velocity characteristic Download PDF

Info

Publication number
CN107024685A
CN107024685A CN201710227554.9A CN201710227554A CN107024685A CN 107024685 A CN107024685 A CN 107024685A CN 201710227554 A CN201710227554 A CN 201710227554A CN 107024685 A CN107024685 A CN 107024685A
Authority
CN
China
Prior art keywords
gesture
gesture identification
convolutional neural
neural networks
time
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN201710227554.9A
Other languages
Chinese (zh)
Inventor
樵明朗
王昊臣
王健伊
王俊
孙忠胜
张松阳
杨伟群
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Beihang University
Original Assignee
Beihang University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Beihang University filed Critical Beihang University
Priority to CN201710227554.9A priority Critical patent/CN107024685A/en
Publication of CN107024685A publication Critical patent/CN107024685A/en
Pending legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S7/00—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00
    • G01S7/02—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00
    • G01S7/41—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00 using analysis of echo signal for target characterisation; Target signature; Target cross-section
    • G01S7/417—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00 using analysis of echo signal for target characterisation; Target signature; Target cross-section involving the use of neural networks
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S7/00—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00
    • G01S7/02—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00
    • G01S7/41—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00 using analysis of echo signal for target characterisation; Target signature; Target cross-section
    • G01S7/415—Identification of targets based on measurements of movement associated with the target
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
    • G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
    • G06V40/107—Static hand or arm

Abstract

A kind of gesture identification method based on apart from velocity characteristic, step is as follows：1, using frequency modulated continuous wave radar as gesture sensor, the Beat Signal received is intercepted and arranged, obtain the dimension matrix of radar echo signal 2；2, to the radar return matrix in step 1, carry out Two-dimensional FFT along fast time dimension and slow time dimension respectively；3, utilize R D graphic sequence training convolutional neural networks；4, by real-time R D sequence inputtings convolutional neural networks model, test is identified, the test accuracy of the convolutional neural networks finally given is more than 98%；5, realize the operation to computer end and the control of material object；The present invention can carry out gesture identification in no illumination and in the case of not dressing sensor, realize with computer end and material object interact；Present invention adds real-time detection, the problem of uncertain identification caused in data cutout position is difficult is cleverly solved, real-time gesture identification and control are realized, with preferable practical value.

Description

A kind of gesture identification method based on distance-velocity characteristic
Technical field
The present invention provides a kind of gesture identification method based on distance-velocity characteristic, be it is a kind of be based on Radar Signal Processing- The gesture identification method of convolutional neural networks, belongs to Radar Technology field and human-computer interaction technique field.
Background technology
It is the precondition that machine understands human gesture's language that automatic identification is carried out to gesture motion.Gesture identification refers to profit Gesture information is gathered with certain sensor, then gesture is identified by mathematical algorithm, its final purpose is to carry out equipment Manipulation and man-machine interaction.With the development of computer technology, gesture identification will turn into after traditional input tool such as mouse, keyboard it Another man-machine interaction mode afterwards.Current gesture identification method, which includes wearing sensor device, optical image recognition etc., to be done Method, but the former needs to dress equipment always in use, it is not only inconvenient, and application is by certain limit System；Optical image recognition method is easily influenceed by environment, and when ambient lighting changes, the error rate of identification can be significantly Increase.
It is a kind of new approach to carry out gesture identification by radar.Radar sends FM electromagnetic ripple to target, receives simultaneously The electromagnetic wave returned is reflected, then by analyzing and processing the signal after JieDuHuaYu II Decoction, distance, speed on target is obtained Etc. information, then pass through the characteristic information that further processing obtains different gestures, reach the purpose of identification gesture.Entered using radar Row gesture identification need not dress numerous and diverse equipment, not influenceed by environmental factors such as illumination, strong antijamming capability, can be round-the-clock Work.Additionally have the advantages that integrated level height, small volume, low in energy consumption, cost be low, real-time is good.
The difference frequency signal that is received to frequency modulated continuous wave radar carries out Two-dimensional FFT (Fast Fourier Transform (FFT)) processing, can be with Obtain range-to-go-velocity information matrix (referred to as R-D figures).The concrete processing procedure of Two-dimensional FFT is：Accumulate first certain The Beat Signal in frequency modulation cycle is simultaneously arranged in matrix, per a line comprising frequency modulation cycle, then respectively to every a line and each Row carry out FFT, that is, obtain the R-D figures of target.This processing method can release the coupling between target velocity and Doppler, carry The precision of high parameter estimation.Different gesture motions has different R-D features, and feature is that can complete gesture identification accordingly.
Convolutional neural networks are that CNN is appeared in the 1960s, Hubel and Wiesel is used in research cat cortex CNN is proposed during the neuron of local sensitivity and set direction.CNN is handled without carrying out complicated early stage to image, passes through office Portion's receptive field and parameter sharing can effectively reduce network parameter number, improve training effectiveness.Depth CNN has multilayer convolution knot Structure, the feature of image can be automatically obtained by convolutional layer, and complete the automatic identification of target.Depth CNN has very strong Habit ability, has become the study hotspot of numerous ambits, extensively should have been obtained in fields such as pattern-recognition, speech analysises With.
The content of the invention
1. purpose：
The purpose of patent of the present invention is to provide a kind of method that gesture identification is carried out based on distance-velocity characteristic.This method By carrying out Two-dimensional FFT conversion to gesture radar return, the R-D figures of target are obtained, in this, as depth CNN input, to not Same gesture is identified.This method can carry out high-precision Real time identification to various gestures, while not by ambient lighting etc. The influence of factor.
2. technical scheme：A kind of gesture identification method based on distance-velocity characteristic
The present invention is a kind of gesture identification method based on distance-velocity characteristic, and contactless hand is carried out using electromagnetic wave Gesture is recognized.By radar emission linear frequency modulation continuous wave signals, target echo signal and JieDuHuaYu II Decoction are received, Beat Signal is obtained, PC ends are sent to after sampled to be handled；Two-dimensional FFT processing is carried out after the Beat Signal of PC ends accumulation certain time length, mesh is obtained Target R-D images；Moving-target detection is carried out according to R-D images, detects after moving target, R-D sequence inputtings is trained Convolutional neural networks carry out real-time gesture identification.
The present invention is a kind of gesture identification method based on distance-velocity characteristic, and this method specific steps include：
Step 1, it is suitable by the slow time to the Beat Signal that receives using frequency modulated continuous wave radar as gesture sensor Sequence is arranged, and obtains the dimension matrix of radar echo signal 2, often one frequency modulation cycle (slow time) of row correspondence, each column correspondence one Range cell (fast time)；
Step 2, to the radar return matrix in step 1, fast Fourier is carried out along fast time dimension and slow time dimension respectively Conversion is FFT, obtains the R-D figures comprising target range and velocity information；The calculation formula of Two-dimensional FFT such as formula (1),
In formula, X (l, k) is the result of Two-dimensional FFT conversion, NFRepresent the sampling number of slow time dimension, NsRepresent fast time dimension Sampling number, xm(n) n-th of sampling in echo matrix in m-th of frequency modulation cycle is represented；
Step 3, through test statistics, after target enters radar beam, significant change can occur for echo signal intensity, accordingly Change can detect that moving target whether there is；Moving-target detection threshold value is set, when echo strength exceedes detection threshold value, i.e., Think that moving target is present；Detect after moving target, the frame Beat Signal of system continuous acquisition 12 simultaneously carries out Two-dimensional FFT conversion, Obtain R-D graphic sequences；
Step 4, the R-D sequences for 4 kinds of gestures that step 3 is obtained are randomly selected, are divided into training set and test set；By training set Input depth convolutional neural networks are trained, and test set are identified test with trained convolutional neural networks；Through 10000 repetitive exercises are crossed, the test accuracy of the convolutional neural networks finally given is more than 98%；
Step 5, radar echo signal is gathered in real time by the method for step 3 and detect moving target, moved when detecting gesture R-D sequences are stored and are sent to convolutional neural networks when making and are identified, recognition result are exported in real time, it is possible to achieve to computer The control of the operation at end either material object.
Wherein, it is described " to the radar return matrix in step 1, respectively along fast time dimension and slow time dimension in step 2 It is FFT " to carry out Fast Fourier Transform (FFT), and its practice is as follows：First, FFT (fast time dimension) is carried out to each row of echo matrix Calculate, then carrying out FFT (slow time dimension) to each row successively is calculated, and finally each element modulus to whole matrix is worth to The result of Two-dimensional FFT.
Wherein, it is described in step 3 " to detect after moving target, the frame Beat Signal of system continuous acquisition 12 is simultaneously carried out Two-dimensional FFT is converted, and obtains R-D graphic sequences ", its practice is as follows：Pair R-D matrixes " general power " can be counted in program every time, i.e., The result of whole R-D matrixes modulus value summation, then calculates the difference with last time R-D matrixes " general power ", and if more than one Determine threshold values, be considered as action, then the frame data of continuous acquisition 12, R-D sequences are obtained after carrying out Two-dimensional FFT change respectively Row, if being not above threshold values, gathered data, does not continue to detect.
Wherein, it is described in step 4 " to be trained training set input depth convolutional neural networks, with by training Convolutional neural networks test set is identified test ", its practice is as follows：By the data collected by 4:1 ratio is divided into Training set and test set, add label to every group of gesture data in training set, are then fed into convolutional neural networks and are trained, together When the training that often completes on a training set its prediction accuracy is just detected with test set, if the degree of accuracy is not up to requirement after Continuous training, no person's deconditioning.
Wherein, it is described in steps of 5 " to export recognition result in real time, it is possible to achieve to the operation either reality of computer end The control of thing ", refers to that recognition result is converted into control signal by us, realizes the control to computer end, including browse webpage, The function such as consulted a map, and we are also achieved to the control in kind such as mechanical arm, realize a kind of new man-machine interaction mode.
By above step, we realize the gesture identification not against camera and sensor device, can not have Illumination and gesture identification is carried out in the case of not dressing sensor, and realize and interacted with computer end and material object, the present invention A kind of reliable, novel man-machine mutual interface can be used as.Present invention adds real-time detection, data cutout position is cleverly solved The problem of uncertain identification caused is difficult is put, real-time gesture identification and control are realized, with preferable practical value.
3rd, advantage and effect：
The present invention is a kind of contactless gesture identification method, with following advantage：
(1) present invention recognizes gesture using frequency modulated continuous wave radar, it is to avoid the influence of the factor to identification such as ambient lighting, Improve the reliability that identifying system works under varying environment situation so that man-machine interaction more stablize and conveniently.
(2) different gestures are identified by analyzing distance-velocity characteristic of radar signal by the present invention, are passed through simultaneously 12 frame R-D sequences describe a gesture motion, improve the discrimination of different gestures.
(3) present invention carries out gesture identification using depth convolutional neural networks, can learn between different gesture motions Fine difference, improve the accuracy of gesture identification.
(4) the present invention is kept away using moving object detection and the method for detecting the frame R-D sequences of continuous acquisition 12 after action Having exempted from continuous collecting R-D figures caused may act aliasing, and accuracy rate is improved again, high-precision gesture is realized and detects in real time And identification.
Brief description of the drawings
Fig. 1 is gesture identification method FB(flow block) of the present invention.
Fig. 2 is 4 kinds of gesture schematic diagrames in the embodiment of the present invention.
Embodiment
With reference to the accompanying drawings and examples, technical scheme is described further.
The present invention is a kind of method that gesture identification is carried out based on distance-velocity characteristic, identification process as shown in figure 1, tool Body includes：
Step 1, Modulation Continuous Wave Radar transmitting carrier frequency is 23GHZ signals, and radar antenna beam angle is 12 ° of x 25 °, whole palm can be irradiated.Closely applied because gesture identification belongs to, palm is general within 0.5m apart from radar, institute To set the transmit power of radar as 6dBm.The radar Beat Signal received is arranged by the order of slow time, obtained Radar echo signal 2 ties up matrix, often one frequency modulation cycle (slow time) of row correspondence, and each column one range cell of correspondence is (when fast Between).The radar frequency modulation cycle is 2.5ms, and sample rate is 48kHz, and each radar return matrix includes 32 frequency modulation cycles,.
Step 2, to the radar return matrix in step 1, FFT is carried out along fast time dimension and slow time dimension respectively, is wrapped R-D figures containing target range and velocity information.When an object is moving, R-D figures can change in distance and speed, thus may be used To be used as the feature for recognizing different gestures.The calculation formula of Two-dimensional FFT such as formula (1),
In formula, X (l, k) is the result of Two-dimensional FFT conversion, NFRepresent the sampling number of slow time dimension, NsRepresent fast time dimension Sampling number, xm(n) n-th of sampling in echo matrix in m-th of frequency modulation cycle is represented；
Step 3, through test statistics, after target enters radar beam, significant change can occur for echo signal intensity, accordingly Change can detect that moving target whether there is.To the signal intensity progress summation process after Two-dimensional FFT is converted, and according to Moving-target detection threshold value is set according to test statistics result, when echo strength and during less than threshold value, it is believed that without gesture motion, give up This frame data, into the collection and processing of next frame data, so will not both take memory space, it also avoid static position The influence of lower noise；When echo strength and during more than detection threshold value, that is, think that moving target is present, the frame of system continuous acquisition 12 is poor Signal is clapped, Two-dimensional FFT conversion is carried out, obtains R-D graphic sequences and store, is used for CNN training and test.4 kinds of gestures of test As shown in Fig. 2 the duration of each gesture is within 1s here, the whole gesture of representative that 12 frame data can be complete Action process.R-D sequences are to avoid the aliasing caused by duration data collection between fore-aft motion, and basic guarantee action The integrality of information, realizes high-precision real-time gesture detection and recognizes.The corresponding R-D graphic sequences of every kind of gesture are all not Together, such as the R-D graphic sequences of preceding pushing hands gesture have the process that an obvious speed diminishes after first becoming greatly, and rotate the R-D of gesture Graphic sequence peak value can all occur on positive-negative velocity direction.Because every kind of gesture has specific R-D sequence signatures, therefore, according to It is feasible that this feature, which carries out gesture identification,.
Step 4, the R-D sequences that read step 3 is stored, randomly select 80% data as training set, remaining data As test set, depth convolutional neural networks are trained and tested.Three people are acquired altogether during generation training test data Gesture data, everyone each gesture motion repeats 50 times.So the training set of each gesture includes 120 groups of R-D sequences, test Collection includes 30 groups of R-D sequences, and the recognition accuracy of test set has reached more than 98%.
Step 5, gathered in real time and detection radar echo signal by the method for step 3, when detecting gesture motion by number The depth convolutional neural networks trained according to storing and being sent to are identified, and recognition result is exported in real time.Each gesture motion It is repeated 20 times, real-time testing result is as shown in table 1.
Table 1：4 class gesture identification success rate tables
Gesture/accuracy rate It is static Before push away Post-tensioning Overturn palm
It is static 95% 0 0 5%
Before push away 0 100% 0 0
Post-tensioning 0 0 100% 0
Overturn palm 5% 0 0 95%
From experimental result it can be seen that, the Real time identification accuracy rate of each gesture motion is very high, has reached more than 95%, Wherein before push away the discrimination acted with post-tensioning and reach 100%.Its reason is：1st, frequency modulated continuous wave radar progress is employed Close-in measurement, the precision of ranging and range rate is very high；2nd, by the use of the R-D figures of gesture as identification feature, and 12 frame sequences are used As the complete description of a gesture, distance and length velocity relation in R-D sequences can effectively distinguish different gestures；3rd, employ Depth convolutional neural networks are identified, and the fine feature in various gesture R-D figures can be fully extracted, exactly to different hands Gesture is identified.
In summary, distance-velocity characteristic of the invention by analyzing gesture, utilizes R-D sequence pair depth convolutional Neurals Network is trained, and the network model finally given can carry out real-time, high-precision gesture identification.

Claims (5)

1. a kind of gesture identification method based on distance-velocity characteristic, it is characterised in that：This method specific steps include：
Step 1, using frequency modulated continuous wave radar as gesture sensor, the Beat Signal received is entered by the order of slow time Row arrangement, obtains the dimension matrix of radar echo signal 2, and often row one frequency modulation cycle of correspondence is slow time, each column one distance of correspondence Unit is the fast time；
Step 2, to the radar return matrix in step 1, Fast Fourier Transform (FFT) is carried out along fast time dimension and slow time dimension respectively That is FFT, obtains the R-D figures comprising target range and velocity information；The calculation formula of Two-dimensional FFT such as formula (1),
In formula, X (l, k) is the result of Two-dimensional FFT conversion, NFRepresent the sampling number of slow time dimension, NsRepresent adopting for fast time dimension Number of samples, xm(n) n-th of sampling in echo matrix in m-th of frequency modulation cycle is represented；
Step 3, through test statistics, after target enters radar beam, significant change can occur for echo signal intensity, change accordingly It can detect that moving target whether there is；Moving-target detection threshold value is set, when echo strength exceedes detection threshold value, that is, is thought Moving target is present；Detect after moving target, the frame Beat Signal of system continuous acquisition 12 simultaneously carries out Two-dimensional FFT conversion, obtains R-D graphic sequences；
Step 4, the R-D sequences for 4 kinds of gestures that step 3 is obtained are randomly selected, are divided into training set and test set；Training set is inputted Depth convolutional neural networks are trained, and test set are identified test with trained convolutional neural networks；By 10000 repetitive exercises, the test accuracy of the convolutional neural networks finally given is more than 98%；
Step 5, radar echo signal is gathered in real time by the method for step 3 and detect moving target, when detecting gesture motion R-D sequences are stored and are sent to convolutional neural networks and are identified, recognition result is exported in real time, the behaviour to computer end can be realized Make and control in kind；
By above step, we realize the gesture identification not against camera and sensor device, can in no illumination and Carry out gesture identification in the case of not dressing sensor, and realize and interacted with computer end and material object, can as one kind Man-machine mutual interface lean on, novel；Present invention adds real-time detection, cleverly solve data cutout position and do not know to cause Identification it is difficult the problem of, real-time gesture identification and control are realized, with good practical value.
2. a kind of gesture identification method based on distance-velocity characteristic according to claim 1, it is characterised in that：
It is described in step 2 " to the radar return matrix in step 1, to be carried out respectively along fast time dimension and slow time dimension quick Fourier transformation is FFT ", and its practice is as follows：First, each row progress FFT to echo matrix is that fast time dimension is calculated, then It is that slow time dimension is calculated that each row are carried out with FFT successively, and finally each element modulus to whole matrix is worth to Two-dimensional FFT As a result.
3. a kind of gesture identification method based on distance-velocity characteristic according to claim 1, it is characterised in that：
It is described in step 3 " to detect after moving target, the frame Beat Signal of system continuous acquisition 12 simultaneously carries out Two-dimensional FFT change Change, obtain R-D graphic sequences ", its practice is as follows：R-D matrixes " general power " can be counted in program every time, i.e., to whole R-D squares The result of battle array modulus value summation, then calculates the difference with last time R-D matrixes " general power ", if more than certain threshold values, just Think there is action, then the frame data of continuous acquisition 12, R-D sequences are obtained after carrying out Two-dimensional FFT change respectively, if do not had More than threshold values, then not gathered data, continues to detect.
4. a kind of gesture identification method based on distance-velocity characteristic according to claim 1, it is characterised in that：
It is described in step 4 " training set input depth convolutional neural networks to be trained, trained convolutional Neural is used Network test set is identified test ", its practice is as follows：By the data collected by 4:1 ratio is divided into training set and survey Examination collection, adds label to every group of gesture data in training set, is then fed into convolutional neural networks and is trained, while often completing one Training on secondary training set just detects its prediction accuracy with test set, no if the degree of accuracy is not up to requirement and continues to training Person's deconditioning.
5. a kind of gesture identification method based on distance-velocity characteristic according to claim 1, it is characterised in that：
Described " exporting recognition result in real time, can realize the operation to computer end and the control of material object ", refers to me in steps of 5 Recognition result is converted into control signal, realize the control to computer end, including browse webpage and the function that consults a map, I Also achieve the control in kind to mechanical arm, realize a kind of new man-machine interaction mode.
CN201710227554.9A 2017-04-10 2017-04-10 A kind of gesture identification method based on apart from velocity characteristic Pending CN107024685A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201710227554.9A CN107024685A (en) 2017-04-10 2017-04-10 A kind of gesture identification method based on apart from velocity characteristic

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201710227554.9A CN107024685A (en) 2017-04-10 2017-04-10 A kind of gesture identification method based on apart from velocity characteristic

Publications (1)

Publication Number Publication Date
CN107024685A true CN107024685A (en) 2017-08-08

Family

ID=59526796

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201710227554.9A Pending CN107024685A (en) 2017-04-10 2017-04-10 A kind of gesture identification method based on apart from velocity characteristic

Country Status (1)

Country Link
CN (1) CN107024685A (en)

Cited By (34)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108200706A (en) * 2018-01-25 2018-06-22 宁波隔空智能科技有限公司 A kind of illuminator and its control method based on microwave radar Gesture Recognition
CN108331114A (en) * 2018-01-25 2018-07-27 宁波隔空智能科技有限公司 A kind of closestool and its control method based on microwave radar Gesture Recognition
CN108344995A (en) * 2018-01-25 2018-07-31 宁波隔空智能科技有限公司 A kind of gesture identifying device and gesture identification method based on microwave radar technology
CN108344996A (en) * 2018-01-25 2018-07-31 隔空（上海）智能科技有限公司 A kind of panel-switch and its control method based on microwave radar Gesture Recognition
CN108361780A (en) * 2018-01-25 2018-08-03 宁波隔空智能科技有限公司 Cooker hood controller based on microwave radar Gesture Recognition and control method
CN108828548A (en) * 2018-06-26 2018-11-16 重庆邮电大学 A kind of three Parameter fusion data set construction methods based on fmcw radar
CN108844186A (en) * 2018-03-30 2018-11-20 广东美的制冷设备有限公司 Control method, air conditioner and the computer readable storage medium of air conditioner
CN109164915A (en) * 2018-08-17 2019-01-08 湖南时变通讯科技有限公司 A kind of gesture identification method, device, system and equipment
CN109188414A (en) * 2018-09-12 2019-01-11 北京工业大学 A kind of gesture motion detection method based on millimetre-wave radar
CN109271838A (en) * 2018-07-19 2019-01-25 重庆邮电大学 A kind of three parameter attributes fusion gesture identification method based on fmcw radar
CN109583436A (en) * 2019-01-29 2019-04-05 杭州朗阳科技有限公司 A kind of gesture recognition system based on millimetre-wave radar
CN109829509A (en) * 2019-02-26 2019-05-31 重庆邮电大学 Radar gesture identification method based on fused neural network
CN110031827A (en) * 2019-04-15 2019-07-19 吉林大学 A kind of gesture identification method based on ultrasonic distance measurement principle
CN110111651A (en) * 2018-02-01 2019-08-09 周玮 Intelligent language interactive system based on posture perception
CN110210330A (en) * 2019-05-13 2019-09-06 清华大学 Electromagnetic signal recognition methods and device based on Tacit Knowledge structure figures convolutional network
CN110309689A (en) * 2018-03-27 2019-10-08 南京理工大学 The domain Gabor gesture identification detection method based on ULTRA-WIDEBAND RADAR
CN110647803A (en) * 2019-08-09 2020-01-03 深圳大学 Gesture recognition method, system and storage medium
CN110741385A (en) * 2019-06-26 2020-01-31 Oppo广东移动通信有限公司 Gesture recognition method and device and location tracking method and device
CN110799927A (en) * 2018-08-30 2020-02-14 Oppo广东移动通信有限公司 Gesture recognition method, terminal and storage medium
CN110826426A (en) * 2019-10-21 2020-02-21 中国汽车技术研究中心有限公司 Vehicle-mounted gesture recognition test system and test method
CN110941331A (en) * 2018-09-21 2020-03-31 国际商业机器公司 Gesture recognition using 3D millimeter wave radar
CN111476058A (en) * 2019-01-23 2020-07-31 北京邮电大学 Gesture recognition method based on millimeter wave radar
CN111813222A (en) * 2020-07-09 2020-10-23 电子科技大学 Terahertz radar-based fine dynamic gesture recognition method
CN111813223A (en) * 2020-07-09 2020-10-23 电子科技大学 Gesture automatic segmentation and recognition method based on high-resolution radar signals
CN112415510A (en) * 2020-11-05 2021-02-26 深圳大学 Double-station radar gesture recognition method, device and system and storage medium
WO2021068470A1 (en) * 2020-04-09 2021-04-15 浙江大学 Radar signal-based identity and gesture recognition method
CN112965060A (en) * 2021-02-19 2021-06-15 加特兰微电子科技(上海)有限公司 Detection method and device for vital sign parameters and method for detecting physical sign points
WO2021203364A1 (en) * 2020-04-09 2021-10-14 浙江大学 Radar signal processing method for identity recognition
TWI756122B (en) * 2021-04-30 2022-02-21 開酷科技股份有限公司 Distance Doppler Radar Angle Sensing Method and Device
US11474232B2 (en) 2021-03-19 2022-10-18 KaiKuTek Inc. Range doppler angle detection method and range doppler angle detection device
CN116027325A (en) * 2023-03-31 2023-04-28 泛擎科技(深圳)有限公司 Human body action recognition method and device based on intelligent wearable equipment
US11639985B2 (en) 2020-07-02 2023-05-02 International Business Machines Corporation Three-dimensional feature extraction from frequency modulated continuous wave radar signals
CN117519474A (en) * 2023-11-06 2024-02-06 中国人民解放军陆军工程大学 Radar gesture feature acquisition method considering motion priori
CN111476058B (en) * 2019-01-23 2024-05-14 北京邮电大学 Gesture recognition method based on millimeter wave radar

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105740823A (en) * 2016-02-01 2016-07-06 北京高科中天技术股份有限公司 Dynamic gesture trace recognition method based on depth convolution neural network
US20160252607A1 (en) * 2015-02-27 2016-09-01 Texas Instruments Incorporated Gesture Recognition using Frequency Modulated Continuous Wave (FMCW) Radar with Low Angle Resolution
US20160259037A1 (en) * 2015-03-03 2016-09-08 Nvidia Corporation Radar based user interface

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20160252607A1 (en) * 2015-02-27 2016-09-01 Texas Instruments Incorporated Gesture Recognition using Frequency Modulated Continuous Wave (FMCW) Radar with Low Angle Resolution
US20160259037A1 (en) * 2015-03-03 2016-09-08 Nvidia Corporation Radar based user interface
CN105740823A (en) * 2016-02-01 2016-07-06 北京高科中天技术股份有限公司 Dynamic gesture trace recognition method based on depth convolution neural network

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
YOUNGWOOK KIM AND BRIAN TOOMAJIAN: "Hand Gesture Recognition Using Micro-Doppler Signatures With Convolutional Neural Network", 《DIGITAL OBJECT IDENTIFIER 》 *
夏春艳: "1.3.4 训练集和测试集", 《数据挖掘技术与应用》 *
朱学峰 等: "《第十三届中国过程控制年会论文集 过程控制科学技术与应用》", 31 July 2002 *
费元春: "5.2 超宽带目标检测理论以及时延估计算法的研究", 《超宽带雷达理论与技术》 *

Cited By (52)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108331114A (en) * 2018-01-25 2018-07-27 宁波隔空智能科技有限公司 A kind of closestool and its control method based on microwave radar Gesture Recognition
CN108344995A (en) * 2018-01-25 2018-07-31 宁波隔空智能科技有限公司 A kind of gesture identifying device and gesture identification method based on microwave radar technology
CN108344996A (en) * 2018-01-25 2018-07-31 隔空（上海）智能科技有限公司 A kind of panel-switch and its control method based on microwave radar Gesture Recognition
CN108361780A (en) * 2018-01-25 2018-08-03 宁波隔空智能科技有限公司 Cooker hood controller based on microwave radar Gesture Recognition and control method
CN108200706A (en) * 2018-01-25 2018-06-22 宁波隔空智能科技有限公司 A kind of illuminator and its control method based on microwave radar Gesture Recognition
CN110111651A (en) * 2018-02-01 2019-08-09 周玮 Intelligent language interactive system based on posture perception
CN110309689B (en) * 2018-03-27 2023-04-07 南京理工大学 Gabor domain gesture recognition detection method based on ultra-wideband radar
CN110309689A (en) * 2018-03-27 2019-10-08 南京理工大学 The domain Gabor gesture identification detection method based on ULTRA-WIDEBAND RADAR
CN108844186A (en) * 2018-03-30 2018-11-20 广东美的制冷设备有限公司 Control method, air conditioner and the computer readable storage medium of air conditioner
CN108828548A (en) * 2018-06-26 2018-11-16 重庆邮电大学 A kind of three Parameter fusion data set construction methods based on fmcw radar
CN109271838A (en) * 2018-07-19 2019-01-25 重庆邮电大学 A kind of three parameter attributes fusion gesture identification method based on fmcw radar
CN109271838B (en) * 2018-07-19 2020-11-03 重庆邮电大学 FMCW radar-based three-parameter feature fusion gesture recognition method
CN109164915A (en) * 2018-08-17 2019-01-08 湖南时变通讯科技有限公司 A kind of gesture identification method, device, system and equipment
CN109164915B (en) * 2018-08-17 2020-03-17 湖南时变通讯科技有限公司 Gesture recognition method, device, system and equipment
CN110799927B (en) * 2018-08-30 2022-02-15 Oppo广东移动通信有限公司 Gesture recognition method, terminal and storage medium
EP3640674A4 (en) * 2018-08-30 2020-08-26 Guangdong Oppo Mobile Telecommunications Corp., Ltd. Gesture recognition method, terminal, and storage medium
US11061115B2 (en) 2018-08-30 2021-07-13 Guangdong Oppo Mobile Telecommunications Corp., Ltd. Method for gesture recognition, terminal, and storage medium
CN110799927A (en) * 2018-08-30 2020-02-14 Oppo广东移动通信有限公司 Gesture recognition method, terminal and storage medium
CN109188414A (en) * 2018-09-12 2019-01-11 北京工业大学 A kind of gesture motion detection method based on millimetre-wave radar
CN110941331A (en) * 2018-09-21 2020-03-31 国际商业机器公司 Gesture recognition using 3D millimeter wave radar
CN111476058A (en) * 2019-01-23 2020-07-31 北京邮电大学 Gesture recognition method based on millimeter wave radar
CN111476058B (en) * 2019-01-23 2024-05-14 北京邮电大学 Gesture recognition method based on millimeter wave radar
CN109583436B (en) * 2019-01-29 2023-09-12 杭州朗阳科技有限公司 Gesture recognition system based on millimeter wave radar
CN109583436A (en) * 2019-01-29 2019-04-05 杭州朗阳科技有限公司 A kind of gesture recognition system based on millimetre-wave radar
CN109829509B (en) * 2019-02-26 2022-11-22 重庆邮电大学 Radar gesture recognition method based on fusion neural network
CN109829509A (en) * 2019-02-26 2019-05-31 重庆邮电大学 Radar gesture identification method based on fused neural network
CN110031827B (en) * 2019-04-15 2023-02-07 吉林大学 Gesture recognition method based on ultrasonic ranging principle
CN110031827A (en) * 2019-04-15 2019-07-19 吉林大学 A kind of gesture identification method based on ultrasonic distance measurement principle
WO2020228141A1 (en) * 2019-05-13 2020-11-19 清华大学 Electromagnetic signal identification method and device for constructing graph convolutional network on basis of implicit knowledge
CN110210330B (en) * 2019-05-13 2021-05-28 清华大学 Electromagnetic signal identification method and device for constructing graph convolution network based on implicit knowledge
CN110210330A (en) * 2019-05-13 2019-09-06 清华大学 Electromagnetic signal recognition methods and device based on Tacit Knowledge structure figures convolutional network
CN110741385B (en) * 2019-06-26 2023-11-07 Oppo广东移动通信有限公司 Gesture recognition method and device, and positioning tracking method and device
CN110741385A (en) * 2019-06-26 2020-01-31 Oppo广东移动通信有限公司 Gesture recognition method and device and location tracking method and device
CN110647803A (en) * 2019-08-09 2020-01-03 深圳大学 Gesture recognition method, system and storage medium
CN110826426B (en) * 2019-10-21 2022-07-29 中国汽车技术研究中心有限公司 Vehicle-mounted gesture recognition test system and test method
CN110826426A (en) * 2019-10-21 2020-02-21 中国汽车技术研究中心有限公司 Vehicle-mounted gesture recognition test system and test method
US20210318427A1 (en) * 2020-04-09 2021-10-14 Zhejiang University Method for recognizing identity and gesture based on radar signals
WO2021203364A1 (en) * 2020-04-09 2021-10-14 浙江大学 Radar signal processing method for identity recognition
US11947002B2 (en) * 2020-04-09 2024-04-02 Zhejiang University Method for recognizing identity and gesture based on radar signals
WO2021068470A1 (en) * 2020-04-09 2021-04-15 浙江大学 Radar signal-based identity and gesture recognition method
US11639985B2 (en) 2020-07-02 2023-05-02 International Business Machines Corporation Three-dimensional feature extraction from frequency modulated continuous wave radar signals
CN111813222A (en) * 2020-07-09 2020-10-23 电子科技大学 Terahertz radar-based fine dynamic gesture recognition method
CN111813223A (en) * 2020-07-09 2020-10-23 电子科技大学 Gesture automatic segmentation and recognition method based on high-resolution radar signals
CN111813222B (en) * 2020-07-09 2022-03-29 电子科技大学 Terahertz radar-based fine dynamic gesture recognition method
CN111813223B (en) * 2020-07-09 2022-04-22 电子科技大学 Gesture automatic segmentation and recognition method based on high-resolution radar signals
CN112415510A (en) * 2020-11-05 2021-02-26 深圳大学 Double-station radar gesture recognition method, device and system and storage medium
CN112965060A (en) * 2021-02-19 2021-06-15 加特兰微电子科技(上海)有限公司 Detection method and device for vital sign parameters and method for detecting physical sign points
US11474232B2 (en) 2021-03-19 2022-10-18 KaiKuTek Inc. Range doppler angle detection method and range doppler angle detection device
TWI756122B (en) * 2021-04-30 2022-02-21 開酷科技股份有限公司 Distance Doppler Radar Angle Sensing Method and Device
CN116027325A (en) * 2023-03-31 2023-04-28 泛擎科技(深圳)有限公司 Human body action recognition method and device based on intelligent wearable equipment
CN117519474A (en) * 2023-11-06 2024-02-06 中国人民解放军陆军工程大学 Radar gesture feature acquisition method considering motion priori
CN117519474B (en) * 2023-11-06 2024-05-14 中国人民解放军陆军工程大学 Radar gesture feature acquisition method considering motion priori

Similar Documents

Publication Publication Date Title
CN107024685A (en) A kind of gesture identification method based on apart from velocity characteristic
CN109917347A (en) A kind of radar pedestrian detection method based on the sparse reconstruct of time-frequency domain
CN110348288B (en) Gesture recognition method based on 77GHz millimeter wave radar signal
CN102198003B (en) Limb movement detection and evaluation network system and method
Zhao et al. Cubelearn: End-to-end learning for human motion recognition from raw mmwave radar signals
CN110458095A (en) A kind of recognition methods, control method, device and the electronic equipment of effective gesture
CN102243687A (en) Physical education teaching auxiliary system based on motion identification technology and implementation method of physical education teaching auxiliary system
CN104766320A (en) Bernoulli smoothing weak target detection and tracking method under thresholding measuring
Zheng et al. Human posture reconstruction for through-the-wall radar imaging using convolutional neural networks
CN106772299B (en) One kind is based on apart from matched PD radar weak target Dynamic Programming detection method
Al-Wesabi et al. Artificial Intelligence Enabled Apple Leaf Disease Classification for Precision Agriculture.
CN113837131A (en) Multi-scale feature fusion gesture recognition method based on FMCW millimeter wave radar
CN110031827A (en) A kind of gesture identification method based on ultrasonic distance measurement principle
CN112364689A (en) Human body action and identity multi-task identification method based on CNN and radar image
CN116956222A (en) Multi-complexity behavior recognition system and method based on self-adaptive feature extraction
CN111401180A (en) Neural network recognition model training method and device, server and storage medium
CN111353391A (en) Radar interference effect evaluation method and device, electronic equipment and storage medium thereof
Kim et al. A study on 3D human pose estimation using through-wall IR-UWB radar and transformer
Sonny et al. Carry Objects Detection utilizing mmWave Radar Sensor and Ensemble Based Extra Tree Classifier on the Edge Computing Systems
CN110309689B (en) Gabor domain gesture recognition detection method based on ultra-wideband radar
CN115754956A (en) Millimeter wave radar gesture recognition method based on envelope data time sequence
Min et al. Highly-accurate gesture recognition based on ResNet with low-budget data gloves
Jiang et al. Continuous human motion recognition based on FMCW radar and transformer
Abdu-Aguye et al. On the Feasibility of On-body Roaming Models in Human Activity Recognition.
Cho et al. Classification of human body motions using an ultra-wideband pulse radar

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
RJ01 Rejection of invention patent application after publication
RJ01 Rejection of invention patent application after publication

Application publication date: 20170808