CN109947238A - A method of the non-cooperative gesture identification based on WIFI - Google Patents
A method of the non-cooperative gesture identification based on WIFI Download PDFInfo
- Publication number
- CN109947238A CN109947238A CN201910041979.XA CN201910041979A CN109947238A CN 109947238 A CN109947238 A CN 109947238A CN 201910041979 A CN201910041979 A CN 201910041979A CN 109947238 A CN109947238 A CN 109947238A
- Authority
- CN
- China
- Prior art keywords
- data
- wifi
- gesture identification
- feature
- state information
- 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.)
- Granted
Links
Abstract
The method for the non-cooperative gesture identification based on WIFI that the present invention relates to a kind of, solution is the low technical problem of accuracy rate, by using utilize CSI as detection means for solves the problems, such as utilize RSSI it is inadequate to human body gesture identification precision.The present invention uses the higher general USRP of ratio of precision to use timestamp to carry out time synchronization to divide the method for CSI data block to acquire CSI data as launch installation as wireless signals reception equipment and wireless router first, and using Principal Component Analysis and more accurate wavelet transformation removal CSI data noise and extract data characteristics, and feature selecting is carried out using XGBoost algorithm, finally with random forests algorithm come the technical solution to different gesture category classifications, the problem is preferably resolved, can be used in gesture identification.
Description
Technical field
The present invention relates to gesture identification fields, and in particular to a method of the non-cooperative gesture identification based on WIFI.
Background technique
To meet the needs of contemporary people convenient and fast to high speed data service, it is indispensable that WIFI early has become people's life
A technology.According to statistics, 97% is early had reached in the coverage rate of the WIFI of emphasis public arena.Research finds WIFI in addition to energy
It provides outside people's data service, moreover it is possible to utilize the motion conditions of WIFI signal detection human body.Using WIFI signal to human motion
Detection system is broadly divided into two classes: indicating (Received Signal using the intensity of the reception signal of WIFI signal
Strength Indicator, RSSI) it is detection and utilizes channel state information (Channel State using WIFI signal
Information, CSI) it is detected.Since the two Testing index all have very high sensibility to environment, when human body is transported
When dynamic, the Testing index that receiver receives can substantially change the state that movement is judged with this, and RSSI refers to signal in certain a period of time
The average value of interior power indicates the intensity for receiving signal.RSSI, which is easy to obtain, early has become the extensive and mature detection of comparison
Therefore index, however RSSI is merely representative of the average value of signal, the amplitude information containing signal have ignored in signal phase information
The accuracy of detection is not significant enough.CSI refers to the channel state information of signal, compares RSSI, and CSI not only contains reception signal
Amplitude information further comprise its phase information, and be accurate to the channel information of each subcarrier of signal, it is big to compare RSSI
The granularity of numerical value is improved greatly.Therefore the index of the motion detection of current mainstream is mainly according to CSI as motion detection index.
Carrying out motion detection using CSI at present can substantially be divided into: using CSI to position of human body Information locating, and differentiate
Its motion profile.Room area is cut into lattice by some scholars, receives the CSI of signal every to WIFI using those who are investigated
The change degree of the position of a lattice constructs WIFI fingerprint base to match the specific location of those who are investigated;It is imaged using WIFI,
Some scholars are imaged partes corporis humani position using WIFI, form the pattern of a 3D;Test section is carried out to human body gesture using CSI
The human body attitude after wall can be identified by dividing scholar to eliminate interference using mimo antenna system.In the realization for carrying out gesture identification,
Many scholars go to receive WIFI signal using the WIFI equipment of existing business, however can introduce more noises using these equipment
It is low to lead to do gesture identification accuracy rate, some scholars use 2 USRP equipment to go acquisition CSI data, but he as R-T unit
And WIFI agreement is not used, cause the deviation in receiving and transmitting signal in working frequency to cause phase offset in this way, to lead
Cause gesture identification precision very low.In addition, the environment of gesture identification carries out in the ideal case in previous research.
The present invention proposes a kind of non-cooperative gesture identification method based on WIFI signal, not connecting based on WIFI data packet
Continuous property proposes and carries out time synchronization using the method for addition timestamp to divide CSI data block, using PCA algorithm and discrete
Wavelet Transformation Algorithm removes CSI data noise and extracts data characteristics, improves the precision of gesture identification.
Summary of the invention
The technical issues of the technical problem to be solved by the present invention is to gesture identification low precisions existing in the prior art.It mentions
For a kind of method of the non-cooperative gesture identification of new WIFI, the method for the non-cooperative gesture identification of the WIFI has identification
Feature with high accuracy.
In order to solve the above technical problems, the technical solution adopted is as follows:
A method of the non-cooperative gesture identification based on WIFI, which comprises
Step 1, human body make gesture motion, acquire the WIFI data of corresponding characterization channel state information;
Step 2 uses addition timestamp method dividing processing to the WIFI data of step 1, to the number after dividing processing
Validity feature is extracted according to progress noise reduction process, then to the data after noise reduction process;
Step 3 classifies to WIFI data, carries out Feature Selection according to the validity feature of step 2 and establishes model,
Data training is carried out using the method for at least two machine learning;
Step 4 calculates the accuracy of the corresponding model of all machine learning methods, Model Error Analysis is carried out, by error
The smallest model of analysis result is defined as optimum classifier algorithm model;
Step 5 inputs WIFI data to be detected, output detection in the optimum classifier algorithm model in step 4
As a result, completing the non-cooperative gesture identification based on WIFI.
The working principle of the invention: since WIFI router sends the uncontrollable of data speed packet, when receiver receives
The corresponding data packet of each gesture sample can not be directly separated out after data packet from a large amount of data packet.Therefore, we are connecing
Generator terminal is received, to a timestamp is all enclosed in the data packet received every time, then according to timestamp by time at equal intervals
Interior all data packets are classified into the data of a gesture sample.
In above scheme, for optimization, further, the addition timestamp method adds timestamp for the time at equal intervals, obtains
To the data of the correspondence gesture of corresponding period.The data acquisition time of gesture is defined as time at equal intervals, such as 2 seconds one
Secondary, 3 seconds primary, so that the data acquired in the corresponding time are identified.In the present invention, it can be set as 2 second time
Interval, while user being needed to execute gesture using same rate in data acquisition, realize gesture and WIFI data in the time
On synchronization.Then further, the data packet number that router actually per second is sent is different, leads to each gesture sample
Data packet number is unequal.When file is downloaded with maximum rate in the end PC, the data packet number of transmission of network is about 1300
(802.11g protocol router, 100M network bandwidth) per second, we therefrom 1024 data packets of stochastical sampling (because 1024 with
Actual data packet quantity is close, samples available more complete initial data distribution), result in the hand of stabilizing amount
Gesture sample data, convenient for carrying out subsequent signal processing flow.
Further, the noise reduction process is using Principal Component Analysis:
Step 1, the matrix form of the WIFI data of definition characterization channel state information is expressed asH=[H1,H2,...,H52]T;
Wherein, HjIndicate the column vector of channel state information, j=1,2 ..., 52;
Step 2, H is calculatedjMean value, construct mean vector
Step 3, covariance matrix is calculated
Step 4, the feature of covariance matrix is decomposed, calculates C=U Λ UT;
Step 5, the preceding k characteristic value for retaining Λ, reconstructs H-matrix to obtain new channel state information matrix;
Wherein, the selection of k meets condition
Step 6, channel state information matrix is rebuild according to the preceding k characteristic value of step 5 and corresponding feature vector.
Further, the validity feature that extracts is using the method for wavelet transform.
Further, Feature Selection is carried out using the algorithm model of XGBoost to the validity feature of step 2.
Further, the method for the machine learning includes random forests algorithm, and random forests algorithm includes:
Step a, it is therefrom random and put back to and extract k self-service sample sets if original training set data is N;
Step b selects m character subset, each decision tree point to each sample characteristics dimension M from M feature at random
Optimal characteristics are selected from m feature when splitting, k independent decision trees can be established to k sample set;
Step c obtains prediction result according to each k decision tree and votes to obtain final prediction result.
Further, the WIFI data of the corresponding characterization channel state information of the acquisition of step 1 include:
The WIFI data packet of step A, the characterization channel state information of definition are CSI data, using based on software radio
The OFDM receiver of platform GNU Radio, probe data packet examine the OFDM data packet quantity in current demand signal, calculate data
Decision metric m is surveyed in detectiven:There is m when continuousnJudgement detects OFDM data packet when > 0.8, is otherwise judged as OFDM
Data packet number is 0, is detected again;
Step B, clock is synchronous, the initial position of first OFDM symbol is determined, if calculating matched filtering window Ls
Index { the i of interior maximum two matched filtering results1,i2}=argmax (fn), then the position of first long symbols is defined as is
=max (i1,i2)+64;
Step C carries out channel estimation and equalization, extracts the long symbols l of receiver signalk(k=0,1 ... 63) it, finds out
Frequency domain value
Step D records CSI data, calculates CFR value(wherein k=0,1 ..., 63), wherein CSI value is
For the sampled value of CFR;
Wherein,sn+kFor the collected data flow of n+k reception equipment, sn+k+16For delay 16
The data of a sampling, LdFor sampling window;Expression receives the mean power of signal;
Indicate the result of preamble long symbols matched filtering;lkIndicate the sampled value of Preamble long symbols time domain;LkIt indicates
The sampled value of Preamble long symbols frequency domain, Lk=FFT [lk], FFT [] indicates Fourier transformation;Expression connects
The sampled value of receipts machine signal frequency domain.
Beneficial effects of the present invention: the present invention mainly utilizes channel state information CSI as detection means for solving benefit
With the problem that signal strength designation date RSSI is inadequate to human body gesture identification precision.Using the higher common software of ratio of precision without
Line electricity peripheral hardware (Universal Software Radio Peripheral, USRP) is as wireless signals reception equipment and wirelessly
Router acquires CSI data as launch installation.Based on the discontinuity of wireless router transmitting WIFI data packet, the present invention
Time synchronization also is carried out using timestamp to divide the method for CSI data block, and using Principal Component Analysis and more accurate
The methods of wavelet transformation removes CSI data noise and extracts data characteristics, and carries out feature choosing using XGBoost algorithm scheduling algorithm
It selects, finally with machine learning algorithms such as random forests algorithms come to different gesture category classifications.Largely improve ideal shape
The accuracy rate of view distance environment gesture identification reaches 96% under condition, reaches 92% under non line of sight.When occurring certain people in experimental situation
Gesture identification accuracy rate still maintains 90% or more when number interference and WIFI transceiver distance change.
Detailed description of the invention
Present invention will be further explained below with reference to the attached drawings and examples.
Fig. 1, the algorithm flow chart of embodiment 1.
Fig. 2, several algorithms of different of embodiment 1 with feature accuracy rate relational graph.
Fig. 3 carries out feature selecting algorithms of different accuracy rate comparison diagram using XGBoost algorithm.
Fig. 4, the experimental result picture that optimal algorithm obtains in embodiment.
The algorithm of Fig. 5, embodiment show figure under the interference of different numbers.
Fig. 6, performance figure of the algorithm of embodiment under different R-T unit distances.
Specific embodiment
In order to make the objectives, technical solutions, and advantages of the present invention clearer, with reference to embodiments, to the present invention
It is further elaborated.It should be appreciated that described herein, specific examples are only used to explain the present invention, is not used to limit
The fixed present invention.
Embodiment 1
The method for the non-cooperative gesture identification based on WIFI that the present embodiment provides a kind of, such as Fig. 1, which comprises
Step 1, human body make gesture motion, acquire the WIFI data of corresponding characterization channel state information;
Step 2 uses addition timestamp method dividing processing to the WIFI data of step 1, to the number after dividing processing
Validity feature is extracted according to progress noise reduction process, then to the data after noise reduction process;
Step 3 classifies to WIFI data, carries out Feature Selection according to the validity feature of step 2 and establishes model,
Data training is carried out using the method for at least two machine learning;
Step 4 calculates the accuracy of the corresponding model of all machine learning methods, Model Error Analysis is carried out, by error
The smallest model of analysis result is defined as optimum classifier algorithm model;If Fig. 2 is that the concentration that the present embodiment uses is different
The accuracy rate relational graph of algorithm model;
Step 5 inputs WIFI data to be detected, output detection in the optimum classifier algorithm model in step 4
As a result, completing the non-cooperative gesture identification based on WIFI.
Detected person requires to make within the set time on 8 kinds of different gestures are respectively, under, it is left, it is right, it draws a circle, boxes,
Windowing, sliding, by different gestures to the change for the CSI data for receiving signal accordingly to different gesture classifications.
General WIFI data acquisition can be used in the present embodiment.Under preferable case, method realization can be used:
The WIFI data packet of step A, the characterization channel state information of definition are CSI data, using based on software radio
The OFDM receiver of platform GNU Radio, wireless router is as transmitting equipment;Subsequent probe data packet examines current demand signal
In OFDM data packet quantity, calculate Packet probing decision metric mn:There is m when continuousnDetection is judged when > 0.8
To OFDM data packet, otherwise it is judged as that OFDM data packet quantity is 0, detects again;
Step B, clock is synchronous, the initial position of first OFDM symbol is determined, if calculating matched filtering window Ls
Index { the i of interior maximum two matched filtering results1,i2}=argmax (fn), then the position of first long symbols is defined as is
=max (i1,i2)+64;
Step C carries out channel estimation and equalization, extracts the long symbols l of receiver signalk(k=0,1 ... 63) it, finds out
Frequency domain value
Step D obtains CSI data, calculates CFR value(wherein k=0,1 ..., 63), wherein CSI value is
For the sampled value of CFR;
Wherein,sn+kFor the collected data flow of n+k reception equipment, sn+k+16For delay 16
The data of a sampling, LdFor sampling window;Expression receives the mean power of signal;
Indicate the result of preamble long symbols matched filtering;lkIndicate the sampled value of Preamble long symbols time domain;LkIt indicates
The sampled value of Preamble long symbols frequency domain, Lk=FFT [lk], FFT [] indicates Fourier transformation;Expression connects
The sampled value of receipts machine signal frequency domain.
The OFDM receiver of Software Radio platform GNU Radio in the present embodiment, wireless router are set as transmitting
Standby that the prior art can be directly used, this implementation repeats no more.
The extraction validity feature of the present embodiment existing can be mentioned using discrete small wave converting method or using other
Take the algorithm of validity feature.
The algorithm model that the present embodiment can adopt XGBoost to validity feature progress Feature Selection also can be selected other similar
Algorithm model.If Fig. 3 is to carry out feature selecting algorithms of different accuracy rate comparison diagram using XGBoost.
Embodiment 2
The present embodiment is on the basis of embodiment 1, to provide a kind of preferred method for carrying out Noise reducing of data processing, based on
Componential analysis:
Step 1, the matrix form of the WIFI data of definition characterization channel state information is expressed asH=[H1,H2,...,H52]T;
Wherein, HjIndicate the column vector of channel state information, j=1,2 ..., 52;
Step 2, H is calculatedjMean value, construct mean vector
Step 3, covariance matrix is calculated
Step 4, the feature of covariance matrix is decomposed, calculates C=U Λ UT;
Step 5, the preceding k characteristic value for retaining Λ, reconstructs H-matrix to obtain new channel state information matrix;
Wherein, the selection of k meets condition
Step 6, channel state information matrix is rebuild according to the preceding k characteristic value of step 5 and corresponding feature vector.
Embodiment 3
The present embodiment provides a kind of machine learning algorithm that performance is more outstanding on the basis of embodiment 2, is random gloomy
Woods algorithm, random forests algorithm include:
Step a, it is therefrom random and put back to and extract k self-service sample sets if original training set data is N;
Step b selects m character subset, each decision tree point to each sample characteristics dimension M from M feature at random
Optimal characteristics are selected from m feature when splitting, k independent decision trees can be established to k sample set;
Step c obtains prediction result according to each k decision tree and votes to obtain final prediction result.
As Fig. 4 is optimal algorithm be random forests algorithm when obtained experimental result picture.
Present invention is mainly used for improve the low problem of existing Gesture Recognition Algorithm accuracy, it is especially useful in solves environment and changes
When algorithm the lower problem of accuracy of identification.If Fig. 5-Fig. 6, Fig. 5 are that algorithm shows figure under the interference of different numbers.Fig. 6 is to calculate
Performance figure of the method under different R-T unit distances.
Therefore, beneficial effects of the present invention are as follows: effect one, largely improve ideally view distance environment gesture knowledge
Other accuracy rate reaches 96%, reaches 92% under non line of sight;Effect two, when occurring the interference of certain number and WIFI in experimental situation
Gesture identification accuracy rate still maintains 90% or more when transceiver distance changes.
Although the illustrative specific embodiment of the present invention is described above, in order to the technology of the art
Personnel are it will be appreciated that the present invention, but the present invention is not limited only to the range of specific embodiment, to the common skill of the art
For art personnel, as long as long as various change the attached claims limit and determine spirit and scope of the invention in, one
The innovation and creation using present inventive concept are cut in the column of protection.
Claims (7)
1. a kind of method of the non-cooperative gesture identification based on WIFI, it is characterised in that: the described method includes:
Step 1 makes gesture motion, acquires the WIFI data of corresponding characterization channel state information;
Step 2, to the WIFI data of step 1 using addition timestamp method dividing processing, to the data after dividing processing into
Row noise reduction process, then validity feature is extracted to the data after noise reduction process;
Step 3 classifies to WIFI data, carries out Feature Selection according to the validity feature of step 2 and establishes model, uses
The method of at least two machine learning carries out data training;
Step 4 calculates the accuracy of the corresponding model of all machine learning methods, Model Error Analysis is carried out, by error analysis
As a result the smallest model is defined as optimum classifier algorithm model;
Step 5, inputs WIFI data to be detected in the optimum classifier algorithm model in step 4, output test result,
Complete the non-cooperative gesture identification based on WIFI.
2. the method for the non-cooperative gesture identification according to claim 1 based on WIFI, it is characterised in that: the addition
Timestamp method adds timestamp for the time at equal intervals, obtains the data of the correspondence gesture of corresponding period.
3. the method for the non-cooperative gesture identification according to claim 2 based on WIFI, it is characterised in that: the noise reduction
Processing is using Principal Component Analysis:
Step 1, the matrix form of the WIFI data of definition characterization channel state information is expressed asH=[H1,H2,...,H52]T;
Wherein, HjIndicate the column vector of channel state information, j=1,2 ..., 52;
Step 2, H is calculatedjMean value, construct mean vector
Step 3, covariance matrix is calculated
Step 4, the feature of covariance matrix is decomposed, calculates C=U Λ UT;
Step 5, the preceding k characteristic value for retaining Λ, reconstructs H-matrix to obtain new channel state information matrix;
Wherein, the selection of k meets condition
Step 6, channel state information matrix is rebuild according to the preceding k characteristic value of step 5 and corresponding feature vector.
4. the method for the non-cooperative gesture identification according to claim 3 based on WIFI, it is characterised in that: the extraction
Validity feature using wavelet transform method.
5. according to the method for the non-cooperative gesture identification as claimed in claim 3 based on WIFI, it is characterised in that: to step 2
Validity feature carries out Feature Selection using the algorithm model of XGBoost.
6. according to the method for the non-cooperative gesture identification described in claim 4 or 5 based on WIFI, it is characterised in that: the machine
The method of device study includes random forests algorithm, and random forests algorithm includes:
Step a, it is therefrom random and put back to and extract k self-service sample sets if original training set data is N;
Step b selects m character subset, when each decision tree is divided to each sample characteristics dimension M from M feature at random
Optimal characteristics are selected from m feature, k independent decision trees can be established to k sample set;
Step c obtains prediction result according to each k decision tree and votes to obtain final prediction result.
7. according to the method for the non-cooperative gesture identification as claimed in claim 6 based on WIFI, it is characterised in that: step 1 is adopted
Collecting the corresponding WIFI data for characterizing channel state information includes:
The WIFI data packet of step A, the characterization channel state information of definition are CSI data, using based on Software Radio platform
The OFDM receiver of GNU Radio, probe data packet examine the OFDM data packet quantity in current demand signal, calculate data detective
Survey decision metric mn:There is m when continuousnJudgement detects OFDM data packet when > 0.8, is otherwise judged as OFDM data
Packet quantity is 0, is detected again;
Step B, clock is synchronous, the initial position of first OFDM symbol is determined, if calculated in matched filtering window Ls most
Index { the i of two big matched filtering results1,i2}=argmax (fn), then the position of first long symbols is defined as is=
max(i1,i2)+64;
Step C carries out channel estimation and equalization, extracts the long symbols l of receiver signalk(k=0,1 ... 63), find out frequency domain value
Step D records CSI data, calculates CFR value(wherein k=0,1 ..., 63), wherein CSI value is CFR
Sampled value;
Wherein,sn+kFor the collected data flow of n+k reception equipment, sn+k+16It is adopted for 16 for delay
The data of sample, LdFor sampling window;Expression receives the mean power of signal;It indicates
The result of preamble long symbols matched filtering;lkIndicate the sampled value of Preamble long symbols time domain;LkIndicate Preamble long
The sampled value of symbol frequency domain, Lk=FFT [lk], FFT [] indicates Fourier transformation;Indicate receiver signal frequency
The sampled value in domain.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910041979.XA CN109947238B (en) | 2019-01-17 | 2019-01-17 | Non-cooperative gesture recognition method based on WIFI |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910041979.XA CN109947238B (en) | 2019-01-17 | 2019-01-17 | Non-cooperative gesture recognition method based on WIFI |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109947238A true CN109947238A (en) | 2019-06-28 |
CN109947238B CN109947238B (en) | 2020-07-14 |
Family
ID=67006713
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910041979.XA Expired - Fee Related CN109947238B (en) | 2019-01-17 | 2019-01-17 | Non-cooperative gesture recognition method based on WIFI |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109947238B (en) |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110737201A (en) * | 2019-10-11 | 2020-01-31 | 珠海格力电器股份有限公司 | monitoring method, device, storage medium and air conditioner |
CN111142668A (en) * | 2019-12-27 | 2020-05-12 | 中山大学 | Interaction method for positioning and activity gesture joint identification based on Wi-Fi fingerprint |
CN111898568A (en) * | 2020-08-04 | 2020-11-06 | 深圳清华大学研究院 | Gesture recognition method and related equipment |
CN112131972A (en) * | 2020-09-07 | 2020-12-25 | 重庆邮电大学 | Method for recognizing human body behaviors by using WiFi data based on attention mechanism |
CN112333653A (en) * | 2020-09-22 | 2021-02-05 | 西安电子科技大学 | Identity intelligent identification method and system based on WiFi channel state information |
CN112765550A (en) * | 2021-01-20 | 2021-05-07 | 重庆邮电大学 | Target behavior segmentation method based on Wi-Fi channel state information |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20140193030A1 (en) * | 2013-01-09 | 2014-07-10 | Jeremy Burr | Gesture pre-processing of video stream with hold-off period to reduce platform power |
CN104584670A (en) * | 2012-08-23 | 2015-04-29 | 交互数字专利控股公司 | Method and apparatus for performing device-to-device discovery |
CN104615244A (en) * | 2015-01-23 | 2015-05-13 | 深圳大学 | Automatic gesture recognizing method and system |
CN104780023A (en) * | 2015-04-23 | 2015-07-15 | 西安交通大学 | CSI limited feedback method based on timestamp selection |
CN105353881A (en) * | 2015-12-04 | 2016-02-24 | 深圳大学 | Gesture recognition method and system based on RFID (radio frequency identification devices) |
CN105989264A (en) * | 2015-02-02 | 2016-10-05 | 北京中科奥森数据科技有限公司 | Bioassay method and bioassay system for biological characteristics |
CN108805194A (en) * | 2018-06-04 | 2018-11-13 | 上海交通大学 | A kind of hand-written recognition method and system based on WIFI channel state informations |
-
2019
- 2019-01-17 CN CN201910041979.XA patent/CN109947238B/en not_active Expired - Fee Related
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104584670A (en) * | 2012-08-23 | 2015-04-29 | 交互数字专利控股公司 | Method and apparatus for performing device-to-device discovery |
US20140193030A1 (en) * | 2013-01-09 | 2014-07-10 | Jeremy Burr | Gesture pre-processing of video stream with hold-off period to reduce platform power |
CN104615244A (en) * | 2015-01-23 | 2015-05-13 | 深圳大学 | Automatic gesture recognizing method and system |
CN105989264A (en) * | 2015-02-02 | 2016-10-05 | 北京中科奥森数据科技有限公司 | Bioassay method and bioassay system for biological characteristics |
CN104780023A (en) * | 2015-04-23 | 2015-07-15 | 西安交通大学 | CSI limited feedback method based on timestamp selection |
CN105353881A (en) * | 2015-12-04 | 2016-02-24 | 深圳大学 | Gesture recognition method and system based on RFID (radio frequency identification devices) |
CN108805194A (en) * | 2018-06-04 | 2018-11-13 | 上海交通大学 | A kind of hand-written recognition method and system based on WIFI channel state informations |
Non-Patent Citations (2)
Title |
---|
张添: "基于WiFi信道状态信息的手势识别研究", 《万方》 * |
潘浩: "基于WiFi的人体动作识别方法研究", 《中国优秀硕士学位论文全文数据库信息科技辑》 * |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110737201A (en) * | 2019-10-11 | 2020-01-31 | 珠海格力电器股份有限公司 | monitoring method, device, storage medium and air conditioner |
CN111142668A (en) * | 2019-12-27 | 2020-05-12 | 中山大学 | Interaction method for positioning and activity gesture joint identification based on Wi-Fi fingerprint |
CN111142668B (en) * | 2019-12-27 | 2023-04-18 | 中山大学 | Interaction method based on Wi-Fi fingerprint positioning and activity gesture joint recognition |
CN111898568A (en) * | 2020-08-04 | 2020-11-06 | 深圳清华大学研究院 | Gesture recognition method and related equipment |
CN111898568B (en) * | 2020-08-04 | 2023-06-23 | 深圳清华大学研究院 | Gesture recognition method and related equipment |
CN112131972A (en) * | 2020-09-07 | 2020-12-25 | 重庆邮电大学 | Method for recognizing human body behaviors by using WiFi data based on attention mechanism |
CN112333653A (en) * | 2020-09-22 | 2021-02-05 | 西安电子科技大学 | Identity intelligent identification method and system based on WiFi channel state information |
CN112765550A (en) * | 2021-01-20 | 2021-05-07 | 重庆邮电大学 | Target behavior segmentation method based on Wi-Fi channel state information |
Also Published As
Publication number | Publication date |
---|---|
CN109947238B (en) | 2020-07-14 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109947238A (en) | A method of the non-cooperative gesture identification based on WIFI | |
Li et al. | AF-DCGAN: Amplitude feature deep convolutional GAN for fingerprint construction in indoor localization systems | |
WO2016197648A1 (en) | Action detection and recognition method based on wireless signal | |
CN104158611B (en) | Wireless signal Interference Detection system and method based on spectrum analysis | |
CN106131958A (en) | A kind of based on channel condition information with the indoor Passive Location of support vector machine | |
CN107968689A (en) | Perception recognition methods and device based on wireless communication signals | |
JP7150292B2 (en) | Action recognition system and action recognition method | |
US7970542B2 (en) | Method of detecting, locating, and classifying lightning | |
CN106792808A (en) | Los path recognition methods under a kind of indoor environment based on channel condition information | |
CN107490795B (en) | It is a kind of to realize that human motion state knows method for distinguishing by radar | |
CN103596266A (en) | Method, device and system for detecting and locating human body | |
CN105743612B (en) | The method that Real-Time Blind solution tunes up frequency displacement short-term burst signal | |
CN109640269A (en) | Fingerprint positioning method based on CSI Yu Time Domain Fusion algorithm | |
CN110475221A (en) | A kind of personnel's action recognition and location estimation method based on channel state information | |
CN112040400B (en) | Single-site indoor fingerprint positioning method based on MIMO-CSI, storage medium and equipment | |
CN110933628B (en) | Fingerprint indoor positioning method based on twin network | |
CN108717175A (en) | Indoor fingerprint positioning method based on region division and sparse support vector regression | |
CN110543842A (en) | Target motion identification method and system based on Wi-Fi signals | |
CN101764786A (en) | MQAM signal recognition method based on clustering algorithm | |
EP4113174A1 (en) | Estimation device, estimation method, and program | |
Cheng et al. | Device-free human activity recognition based on GMM-HMM using channel state information | |
CN110059612A (en) | A kind of gesture identification method and system that the position based on channel state information is unrelated | |
CN112235215A (en) | Wireless channel detection method, storage medium and terminal equipment | |
CN106170139B (en) | A kind of frequency spectrum detecting method and system | |
CN107451605A (en) | A kind of simple target recognition methods based on channel condition information and SVMs |
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 | ||
GR01 | Patent grant | ||
GR01 | Patent grant | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20200714 Termination date: 20220117 |
|
CF01 | Termination of patent right due to non-payment of annual fee |