CN105573498B - A kind of gesture identification method based on Wi-Fi signal - Google Patents
A kind of gesture identification method based on Wi-Fi signal Download PDFInfo
- Publication number
- CN105573498B CN105573498B CN201510939043.0A CN201510939043A CN105573498B CN 105573498 B CN105573498 B CN 105573498B CN 201510939043 A CN201510939043 A CN 201510939043A CN 105573498 B CN105573498 B CN 105573498B
- Authority
- CN
- China
- Prior art keywords
- gesture
- signal
- data
- hand signal
- template
- 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.)
- Active
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
- G06F3/01—Input arrangements or combined input and output arrangements for interaction between user and computer
- G06F3/017—Gesture based interaction, e.g. based on a set of recognized hand gestures
Landscapes
- Engineering & Computer Science (AREA)
- General Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Human Computer Interaction (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- User Interface Of Digital Computer (AREA)
Abstract
The invention discloses a kind of gesture identification method based on Wi Fi signals, extract hand signal GS to be identified, hand signal GS to be identified is matched with multiple template hand signal, obtain the matching distance between hand signal GS and template hand signal to be identified, the gesture represented by template hand signal wherein corresponding to minimum matching distance is consistent with the gesture that hand signal to be identified represents, effectively avoiding producing substantial amounts of redundant data, either gesture data extraction mistake causes identification mistake or system operation slow.
Description
Technical field
The invention belongs to human-computer interaction, machine learning field, is related to a kind of gesture identification method based on Wi-Fi signal.
Background technology
With the fast development of computer information technology, human-computer interaction technology plays in daily life
More and more important role.Gesture is a kind of people and most intuitively exchange way during extraneous link up, people can by body or
Gesture is directly perceived, succinct, natural terrain reaches the idea of oneself, therefore the human-computer interaction technology based on gesture becomes the heat studied at present
Point, i.e. Gesture Recognition.
Gesture Recognition is broadly divided into two classes at present, and one kind carries special sensor or equipment for target, i.e., main
Dynamic formula Gesture Recognition, active Gesture Recognition mainly carry 3-axis acceleration sensor, gyroscope, electricity by target
The sensor devices such as sub- compass gather hand-type or tracking hand spatial movement data, at present data hand in active gesture identification
Cover it is most widely used, data glove be by target wearing include multiple sensors gloves, by sensor record hand in sky
Between in movement locus and finger-joint movable information, so as to identify the gesture of target, at home, Harbin industry is big
Wujiang qin, high text etc. use the Cyber Glover data being made of 18 sensors in Chinese Sign Language identifying system
Gloves, with reference to Hidden Markov Model (Hidden Markov Model, HMM) and artificial neural network (Artificial
Neural Network, ANN) two methods, realize that to the discrimination of isolated word be 90%, simple statement discrimination is 92%.
Active Gesture Recognition can directly obtain human hand coordinate in space and finger motion information, the accuracy of data
It is high, can recognize that multiple gestures and identification precision is high, but due to also needing user to carry special equipment, be inconvenient to operate, unsuitable
The shortcomings of remote-controlled operation, be limited by very large its application scenarios.
In order to which user can preferably be experienced, the auxiliary devices such as sensor or special equipment are eliminated the reliance on, it is domestic
The outer researcher target that begins one's study does not carry the Gesture Recognition of any sensor or equipment, i.e. passive type gesture identification skill
Art, it is easy to operate since its cost is low, meet the features such as user is accustomed to, become the hot spot studied both at home and abroad.Existing passive type
Gesture Recognition mainly has：Gesture Recognition based on computer vision, the Gesture Recognition based on sound wave, be based on
The Gesture Recognition of Wi-Fi signal.The wherein identification technology based on computer vision, is to catch target gesture by camera
Action sequence, to hand signal carry out complex process, then by pattern matching algorithm carry out gesture identification, be currently based on and regard
The identification technology of frequency is quite ripe, has been successfully applied in daily life, as Microsoft develop 360 game hosts of Xbox,
But since it is easily influenced by ambient light is strong and weak, video processing data amount is big, there are lower deployment cost cost is big, easily reveal hidden
The shortcomings such as private, limit the prospect of its development.Gesture Recognition based on sound wave, most typically successful case are by Hua Sheng
University and Microsoft Research propose jointly, send and connect by the loudspeaker on smart mobile phone or laptop and microphone
Perceived the fluctuation of the gesture of user by 18Khz sound waves, but since the scope that sound wave can perceive is small, limit its application range with
Scene.And nowadays, perfect Wi-Fi infrastructure so that Wi-Fi signal is nearly ubiquitous, and domestic and international scientist utilizes
The universality of Wi-Fi signal, sets about research by the Gesture Recognition centered on perceiving Wi-Fi signal disturbance, as Q.Pu is taught
Award and sent and received using USRP (Universal Software Radio Peripheral, general software radio peripheral hardware)
WI-FI signals, and by Wi-Fi signal carry out OFDM modulation after, analyze the Doppler shift of subcarrier, realize to 9 kinds
Gesture identification.This is can be seen that using the technology of Wi-Fi signal by changing in following high degree from international forward position progress
Kind human lives, improve the quality of life of the mankind.
Gesture identification based on wireless signal has complied with the trend of following man-machine interactive development, has very important research
Meaning and practical value, especially its vast potential for future development, have attracted the interest of a large amount of experts both at home and abroad.Over nearly 3 years, base
It is developed rapidly in the Gesture Recognition of wireless signal, is broadly divided into two research directions：
(1) gesture identification is realized using RFID tag
In 2012, University of Melbourne doctor ParvinAsadzadeh, chased after passive label track using receiver
Track, it is 94% to realize the identification rate of precision to gesture.2013, doctor RasmusKrigslund of Aalborg University, disposed more
The receiver of antenna is tracked passive label, realizes simple 3D gesture identifications.In 2014, Jue doctors Wang of MIT,
RFID tag is dressed by finger, realizes virtual handwritten word, by the phase information of receiver analyzing tags, realizes that word is known
Rate is not 96.8%.University of Washington Bryce Kellogg and doctor VamsiTalla, independent research can be received by mobile phone and marked
The hardware device of information is signed, amplitude envelope is carried out to receiving signal, in the case of low-power consumption, low latency, realizes to 8 kinds
The recognition accuracy of gesture 97%.
(2) gesture identification is realized using Wi-Fi signal
In 2013, the D.Katabi of MIT was taught, and uses USRP (Universal Software Radio
Peripheral, general software radio peripheral hardware) more antennas of connection, mimo system is formed, sends and connects by MIMO technology
By 2.4Ghz Wi-Fi signals, realize to the simple gestures detection in wall behind.The QifanPu professors of University of Washington, modification
USRP-N210 underlying protocols, carry out OFDM technology processing to original signal, by sending 5Gh signals, observe each subcarrier
Doppler shift, realizes the identification to 9 kinds of gestures, and mean accuracy reaches 94%.2014, the Stephan of brother's Dettingen university
Doctor Sigg is identified 11 kinds of gestures using the RSSI value of wireless Wi-Fi signal, its precision is 72%.University of Wisconsin
Pedro doctors Melgarejo, receiving terminal using directional aerial receive AP signals, study under two kinds of scenes to 25 kinds of hands
The identification of gesture, 92% accuracy of identification is realized under the scene of high s/n ratio and 84% knowledge is realized under the scene of low signal-to-noise ratio
Other precision.
Work from the studies above as can be seen that can realize fine-grained gesture identification using RFID tag, but need mesh
Mark carries FRID labels, limits the freedom of user, it is impossible to makes that user is more natural, free human-computer interaction, user experience is non-
It is often poor.Based on Wi-Fi signal gesture identification, due to the generality of WIFI infrastructure devices, while user can be realized without inflexible nothing
Human-computer interaction under the conditions of beam, is favored be subject to domestic and international expert, but domestic and foreign scholars are mainly sent and received by USRP at present
WI-FI signals realize gesture identification, but equally exist certain limitation, since USRP is not belonging to common apparatus, and compare
Costliness, actual deployment cost are high.
The content of the invention
It is in view of the above-mentioned drawbacks of the prior art or insufficient, it is an object of the present invention to propose a kind of based on Wi-Fi letters
Number gesture identification method.
To achieve these goals, the present invention adopts the following technical scheme that：
A kind of gesture identification method based on Wi-Fi signal, specifically includes following steps：
Step 1, transmitting terminal send signal, and user makes various gestures between transmitting terminal and receiving terminal, and signal is produced
Disturbance, two reception antennas of receiving terminal are respectively received signal G1 and G3 after being disturbed by gesture；
Step 2, pre-processes signal G1 and G3, and it is sliding that the pretreatment includes normalized, conjugation processing peace
Processing, obtains pretreated signal S1；
Step 3, hand signal GS to be identified is extracted according to signal S1；
Step 4, builds template hand signal storehouse, multiple template hand signal is included in template hand signal storehouse；It will wait to know
Other hand signal GS is matched with multiple template hand signal respectively, obtains hand signal GS to be identified and multiple template gesture
Matching distance between signal, wherein gesture represented by template hand signal corresponding to minimum matching distance with it is to be identified
The gesture that hand signal represents is consistent.
Specifically, the implementation of the step 2 includes：
Step 2.1:Signal G1 and G3 are normalized, respectively obtain signal G4 and G5 after normalization；
Step 2.2:Conjugation processing is carried out to the signal G4 and G5 after normalization, obtains removing the signal S after noise；
Step 2.3:Signal S after denoising is smoothed, obtains the signal S1 after smoothing processing.
Specifically, the implementation of the step 3 includes：
Step 3.1：According to signal S1, sliding window amplitude and matrix A [n] are calculated, n represents data sample number；
Step 3.2：Determine gesture data constraints；
Step 3.3：Gesture letter to be identified is obtained according to sliding window amplitude and matrix A [n] and gesture data constraints
Number GS.
Specifically, the implementation of the step 4 includes：
Step 4.1：Template hand signal storehouse is built, multiple template hand signal ref is included in template hand signal storehouse
[k], k represent kth kind template gesture letter, and single template hand signal data are expressed as ref (i), and 1≤i≤d, wherein d represent single
The number of the data point of a template gesture data；
Step 4.2：Hand signal GS to be identified is matched with multiple template hand signal respectively, obtains hand to be identified
Matching distance between gesture signal GS and multiple template hand signal, concrete methods of realizing are as follows：
Hand signal data to be identified are GS (j), and 1≤j≤n, wherein n represent of the data point of gesture data to be identified
Number；
If d=n, hand signal GS to be identified and all template hand signals are calculated by Euclidean distance formula
The matching distance value DS [k] of ref [k], k represent kth kind template gesture, the template hand signal corresponding to minimum matching distance
Represented gesture is consistent with the gesture that hand signal to be identified represents；
If d ≠ n, the Distance matrix D (i, j) of d*n, 1≤i≤d, 1≤j≤n are built；Obtained according to Distance matrix D (i, j)
The matching distance value DS [k] of hand signal to be identified and all template hand signals, k represent kth kind template gesture, minimum
With consistent with the gesture that hand signal to be identified represents apart from the gesture represented by corresponding template hand signal.
Specifically, the implementation of the step 3.3 includes：
Step 3.3.1：One section of non-gesture state data is gathered, is averaged as threshold value threshold；
Step 3.3.2：The threshold value threshold extraction hand signals GS obtained according to step 3.3.1；
Step 3.3.2.1:With A [1] for starting point, A [1] represents first data of sliding window amplitude and matrix A, one by one
Judge | A [n] | whether more than threshold value threshold, obtain in A [n] | A [n] | the continuous first segment for being more than threshold value threshold
Data interval A [Tstart]~A [Tend], data interval A [Tstart]~A [Tend] number of samples be num；Judge num whether
In section [Min, Max], wherein, Min is gesture data duration minimum value, and Max is gesture data duration maximum,
If so, 3.3.2.2 is gone to step, if not, going to step 3.3.2.3；
Step 3.3.2.2：Obtain A [Tstart]~A [Tend] data S1 [T in corresponding signal S1start* offset]~
S1[Tend* offset], offset deviates incrementss every time for sliding window, calculates S1 [Tstart* offset]~S1 [Tend*
Offset] in the sum of all data absolute values V；If V is more than minimum threshold min_gesture, then data point S1 is thought
[Tstart* offset]~S1 [Tend* offset] it is gesture data GS, otherwise data point S1 [Tstart* offset]~S1 [Tend*
Offset] it is non-gesture data, with A [Tend+ 1] it is starting point, goes to step 3.3.2.1；
Step 3.3.2.3：When num values are more than Max, data point S1 [Tstart* offset]~S1 [Tend* offset] be
Non- gesture data, with A [Tend+ 1] it is starting point, goes to step 3.3.2.1；When num values are less than Min, continue to A [Tend] after
Data judge one by one | A [n] | whether more than threshold value threshold, obtain the first segment data for being continuously less than threshold value threshold
Section A [Tend+ 1]~A [Tend1], statistics section A [Tend+ 1]~A [Tend1] data amount check be num1；Judging num1 is
It is no to be more than Min, if so, 3.3.2.4 is gone to step, if not, going to step 3.3.2.5；
Step 3.3.2.4：Data interval A [T may determine that according to gesture duration featurestart]~A [Tend1] it is non-
Gesture data, with A [Tend1+ 1] it is starting point, goes to step 3.3.2.1；
Step 3.3.2.5：Continue to A [Tend1] after data judge one by one | A [n] | whether more than threshold value
Threshold, obtains the second segment data interval A [T for being continuously more than threshold value thresholdend1+ 1]~A [Tend2], statistics
Section A [Tend1+ 1]~A [Tend2] data amount check be num2；Judge the sum of num, num1 and num2 whether section [Min,
Max] in, judge data interval A [T if the sum of three numbers are in the sectionstart]~A [Tend2] corresponding to signal S1 in
Data S1 [Tstart* offset]~S1 [Tend2* offset] it is gesture data GS；Otherwise it is non-gesture data, with A [Tend2+1]
For starting point, 3.3.2.1 is gone to step.
Compared with prior art, the present invention has following technique effect：
1st, realize under the conditions of disposing Least-cost and user's any sensor of carrying or other special equipments are not required
Gesture identification so that user carries out human-computer interaction under natural environment.
2nd, method using the present invention correctly extracts hand signal, and hand signal GS to be identified and multiple template gesture are believed
Number matched, obtain the matching distance between hand signal GS and template hand signal to be identified, wherein minimum matching away from
It is consistent with the gesture that hand signal to be identified represents from the gesture represented by corresponding template hand signal.Effectively avoid producing
Either gesture data extraction mistake causes identification mistake or system operation slow to substantial amounts of redundant data.
Brief description of the drawings
Fig. 1 is the algorithm flow chart of the present invention；
Fig. 2 is experiment scene figure；
Fig. 3 is experiment gesture figure；
Fig. 4 is that gesture data carries figure；
Fig. 5 is DTW algorithm search path profiles；
Explanation and illustration in further detail is done to the solution of the present invention with reference to the accompanying drawings and examples.
Embodiment
Above-mentioned technical proposal is deferred to, the gesture identification method of the invention based on Wi-Fi signal, specifically includes following step
Suddenly：
Step 1：Transmitting terminal sends signal, and user makes various gestures between transmitting terminal and receiving terminal, and signal is produced
Disturbance, two reception antennas of receiving terminal are respectively received signal G1 and G3 after being disturbed by gesture.
Signal is sent using existing Wi-Fi infrastructure devices, Wi-Fi infrastructure devices use router, area of the user at 4*4 meters
Five kinds of gestures are done in domain, wherein five kinds of gestures are respectively one hand is pushed forward, is pushed away before and after front jumping, both hands, left hand draws circle, both hands open, are used
Different gestures are done at family, which in the region, to produce different disturbance to the Wi-Fi signal that Wi-Fi infrastructure devices produce, by with it is portable
The Wi-Fi signal that the Inter5300NIC network interface card receiving routers that formula computer is connected are sent, Inter5300NIC network interface cards have three
Root reception antenna, the Wi-Fi signal that connects being respectively received is G1, G2, G3, as shown in figure.The deployment facility it is not only general and
And it is cheap, have good market prospects, such as somatic sensation television game, smart home.
Step 2, signal G1 and G3 be subject to multipath, noise and external environment due to being influenced, it is therefore desirable to which signal is carried out
Pretreatment, obtains pretreated signal S1, specifically includes following steps：
Step 2.1：Due to personal habits and the difference of surrounding environment, cause at different moments same gesture motion can produce difference
It is different, therefore by the way that signal G1 and G3 are normalized, to reduce space-time influence, improve the robustness of the method for the present invention.
Concrete methods of realizing is as follows：
Wherein, G1 [i] represents i-th of data of signal G1, and G4 represents the signal after G1 normalization；G3 [i] represents signal
I-th of data of G3, G5 represent the signal after G3 normalization.
Step 2.2：Since influence of the multi-path jamming to three antennas of receiving terminal is identical, the present invention utilizes common mode inhibition capacity
Noise is eliminated, the signal that wherein antenna 1 and antenna 3 receive includes hand signal and multipath signal.So herein to 1 He of antenna
Signal after the normalization received of antenna 3 carries out conjugation processing, obtains removing the signal S after noise, specific formula for calculation
It is as follows：
S [i]=G4 [i] * conj (G5 [i]) (3)
Step 2.3:, can be to following Classification and Identifications and gesture number because the signal S after denoising also includes some accidental datas
Large error is caused according to extraction, needs further to be smoothed the signal S after denoising for this, for eliminating burr data
Influence, improve the authenticity of gesture data.Using 5 points three times Least square smooth filtering signal S is smoothed, obtain
Signal S1 after to smoothing processing, concrete methods of realizing are as follows：
Wherein, x1Represent the 1st data in signal S, x2Represent the 2nd data in signal S, xiRepresent in signal S i-th
Data, 3≤i≤m；yiRepresent i-th of data in the signal S1 after smoothing processing, 1≤i≤m.
Step 3, the signal S1 obtained according to step 2 extract hand signal GS to be identified.
Since when unpredictable user makes gesture motion, therefore need the start point data and endpoint data to hand signal
It is detected, the purpose is to detect hand signal from the Wi-Fi signal received, accurate end-point detection is following characteristics
Extraction and pattern-recognition provide accurate data, while redundant data is eliminated, reduce the complexity of pattern match with
The time is calculated, improves the identification precision of system.In practical applications, we limit a gesture motion need to include it is one short
Temporary steady gesture, and after gesture motion terminates, then carry out a steady gesture.The present invention is measured by sliding window
Average amplitude size, given threshold carry out the detection of start point data and endpoint data according to threshold value, and specific method is as follows：
Step 3.1：Calculate sliding window amplitude and matrix
The signal S1 obtained for step 2, sliding window amplitude and matrix A are calculated using sliding window：
Wherein,A [n] be by sliding window after, data sample number be n sliding window
Mouth amplitude and matrix,N A can be obtained by being calculated by sliding window, and wherein win is slip
The size of window, offset deviate incrementss every time for sliding window；Win=4 is worked as by experimental verification, effect during offset=2
Most preferably.Amplitude figure after Fig. 5 is handled for initial data by sliding window.
Step 3.2：According to general gesture feature, gesture data meets following constraints：
TLast∈[Min,Max] (6)
Wherein TlastFor the gesture data duration, and Min=6, Max=9.
Step 3.3：Hand signal GS is obtained according to sliding window amplitude and matrix A and gesture data constraints.
Step 3.3.1:(i.e. user does not do any one section of non-gesture state data of collection user in Wi-Fi regions in advance
The data collected under gesture), it is averaged as threshold value threshold, threshold is obtained equal to 0.2 by experiment；
Step 3.3.2：The threshold value threshold extraction hand signals GS obtained according to step 3.3.1
Step 3.3.2.1:With A [1] for starting point, A [1] represents first data of sliding window amplitude and matrix A, one by one
Judge | A [n] | whether more than threshold value threshold, obtain in A [n] | A [n] | the continuous first segment for being more than threshold value threshold
Data interval A [Tstart]~A [Tend], wherein, TstartRepresent first continuous data point for being more than threshold value threshold, Tend
Represent last continuous data point for being more than threshold value threshold；Data interval A [Tstart]~A [Tend] number of samples be
num；Num is judged whether in section [Min=6, Max=9], if so, 3.3.2.2 is gone to step, if not, going to step
3.3.2.3；
Step 3.3.2.2：Obtain A [Tstart]~A [Tend] data S1 [T in corresponding signal S1start* offset]~
S1[Tend* offset], offset deviates incrementss every time for sliding window, calculates S1 [Tstart* offset]~S1 [Tend*
Offset] in the sum of all data absolute values V；If V is more than minimum threshold min_gesture, then data point S1 is thought
[Tstart* offset]~S1 [Tend* offset] it is gesture data GS, otherwise data point S1 [Tstart* offset]~S1 [Tend*
Offset] it is non-gesture data, with A [Tend+ 1] it is starting point, goes to step 3.3.2.1；
Step 3.3.2.3：When num values are more than Max, data point S1 [Tstart* offset]~S1 [Tend* offset] be
Non- gesture data, with A [Tend+ 1] it is starting point, goes to step 3.3.2.1；When num values are less than Min, due to a complete gesture
Signal may include the data less than threshold value, therefore need to continue to A [Tend] after data judge one by one | A [n] | whether be more than
Threshold value threshold, obtains the first segment data interval A [T for being continuously less than threshold value thresholdend+ 1]~A [Tend1], statistics
Data interval A [Tend+ 1]~A [Tend1] data amount check be num1；Judge whether num1 is more than Min, if so, going to step
3.3.2.4, if not, going to step 3.3.2.5；
Step 3.3.2.4：Data interval A [T may determine that according to general gesture duration featurestart]~A [Tend1]
For non-gesture data, with A [Tend1+ 1] it is starting point, goes to step 3.3.2.1；
Step 3.3.2.5：Continue to A [Tend1] after data judge one by one | A [n] | whether more than threshold value
Threshold, obtains the second segment data interval A [T for being continuously more than threshold value thresholdend1+ 1]~A [Tend2], statistics
Section A [Tend1+ 1]~A [Tend2] data amount check be num2；Judge the sum of num, num1 and num2 whether in section [Min=
6, Max=9] in, data interval A [T are judged if the sum of three numbers are in the sectionstart]~A [Tend2] corresponding to signal
Data S1 [T in S1start* offset]~S1 [Tend2* offset] it is gesture data GS；Otherwise it is non-gesture data, with A
[Tend2+ 1] it is starting point, goes to step 3.3.2.1.
Such as left hand draws circle gesture, shown in Fig. 4, when carrying out endpoint extraction, what is outlined in figure is the hand detected
Gesture acts, and is progressively judged according to constraints, and gesture 1 is because the duration is short and amplitude is less than minimum gesture threshold value
Min_gesture, therefore weed out, gesture 2 is because amplitude is less than minimum gesture threshold value min_gesture therefore weeds out.Gesture is moved
Made for 4 duration more than the maximum gesture duration, therefore be determined as noise signal.Finally identify that gesture motion 3 is real
Gesture data.
Step 4, builds template hand signal storehouse, multiple template hand signal is included in template hand signal storehouse；It will wait to know
Other hand signal GS is matched with multiple template hand signal respectively, obtains hand signal GS to be identified and multiple template gesture
Matching distance between signal, wherein gesture represented by template hand signal corresponding to minimum matching distance with it is to be identified
The gesture that hand signal represents is consistent.
Identification classification is by recognizer, and the gesture data collected and pre-defined template gesture data are carried out
The matching analysis, compares the similarity between the template gesture data for drawing and collecting, finally identifies specific gesture.Due to gesture
There are spatio-temporal difference, i.e., same user repeats to do same gesture motion or different user does same gesture motion
When, since the length of arm, gesture motion speed are different, cause gesture amplitude size and have very big difference on the duration
It is different.And dynamic time warping (Dynamic Time Warping, DTW), be mainly used for solve speech recognition in speech rate not
The problem of same, according to this function, can solve that measurement module in the system is different with sample form time span to ask with the algorithm
Topic.The thought that DTW algorithms are based on Dynamic Programming (Dynamic Programming, DP) solves test gesture data and sample hand
Gesture data time template matches problem different in size, builds distance matrix by calculating Euclidean distance, calculates shortest path
Distance and, specific gesture is finally identified according to minimum range.It is as follows to implement step：
Step 4.1：Template hand signal storehouse is built, multiple template hand signal is included in template hand signal storehouse；
User first does 5 kinds of gesture motions in Wi-Fi regions, the gesture data for then producing this 5 kinds of gesture motions
Manual extraction comes out, and is stored as template hand signal ref [k], and k represents kth kind template hand signal.Single template hand
Gesture signal data is expressed as ref (i), and 1≤i≤d, wherein d represent the number of the data point of template gesture data.
Step 4.2：Hand signal GS to be identified is matched with multiple template hand signal respectively, obtains hand to be identified
Matching distance between gesture signal GS and multiple template hand signal.Concrete methods of realizing is as follows：
Hand signal data to be identified are GS (j), and 1≤j≤n, wherein n represent of the data point of gesture data to be identified
Number.
If d=n, hand signal GS and template hand signal ref to be identified is calculated by Euclidean distance formula
Matching distance DS between [k]：
Wherein i=[1, d] j=[1, n], DS represent the distance value of gesture data and template data.Obtain gesture to be identified
The matching distance value DS [k] of signal and all template hand signals, k represent kth kind template hand signal, minimum matching distance
Gesture represented by corresponding template hand signal is consistent with the gesture that hand signal to be identified represents.
If d ≠ n, need ref (i) and GS (j) mapping alignment, DTW algorithms using the thought of Dynamic Programming (DP), will
Gesture data to be identified represents that on the transverse axis of rectangular coordinate system template gesture data is represented on the longitudinal axis of rectangular coordinate system,
Form the grid matrix of a d*n；Calculate each of each data of template gesture data and gesture data to be identified
The distance between data, build the Distance matrix D (i, j) of d*n, and calculation formula is as follows：
D (i, j)=(ref [i]-GS (j))2 (8)
Wherein, 1≤i≤d, 1≤j≤n.
Matching distance DS is obtained according to Distance matrix D (i, j), path must be put from the D (1,1) of two-dimensional coordinate system to be started,
Terminate in (d, n) point.
DS (1,1)=D (1,1) is made, calculates the value of first row DS (d, 1), calculation formula is as follows：
DS (i, 1)=DS (i-1,1)+D (i, 1) wherein 2≤i≤d (8)
The value of the first row DS (1, n) is calculated, calculation formula is as follows：
DS (1, j)=DS (1, j-1)+D (1, j) wherein 2≤j≤n (9)
Constraints is selected to calculate DS (i, j) according to shortest path：
DS (i, j)=D (i, j)+min [DS (i-1, j), DS (i-1, j-1), DS (i, j-1)] (10)
Wherein, 2≤i≤d, 2≤j≤n.
It is matching distance to finally obtain DS (d, n), and DTW searching routes are as shown in Figure 5.Obtain gesture data and all templates
The matching distance value DS [k] of data, k represent kth kind template gesture, the template hand signal institute corresponding to minimum matching distance
The gesture of expression is consistent with the gesture that hand signal to be identified represents.
Experimental verification
This sample plot point is selected in Northwest University's Information Institute Stall, and the size for testing selected areas is 4*4 meters, and user stands
Gesture interaction is carried out between in the zone.For existing wlan device, (TP-LINK routers (AP), have this experiment porch transmitting terminal
Two dual-mode antennas), receiving terminal receives the CSI data from AP using 5300 network interface cards of Intel (having three reception antennas), connects
Receiving end is run in the case where operating system is Ubuntu 10.04.4, and experiment scene is as shown in Figure 2.This experiment gathers five kinds of gestures altogether
This method is verified in action, wherein five kinds of gestures are respectively one hand is pushed forward, is pushed away before and after front jumping, both hands, left hand draws circle, both hands
Open, gesture motion is as shown in Figure 3.We collect altogether 250 groups of data to every kind of 50 groups of data of gesture repeated acquisition, its
In we choose 20 groups of gesture datas as gesture template refer to, 30 groups of residue is as gesture data to be identified.Build gesture
Behind priori storehouse, method is verified, recognition result is as shown in table 1：
Table 1
Ranks represent test data and template data respectively in table, as in the first row 28 expression one hands be pushed forward test data and
One hand is pushed forward template data successful match 28 times, and 2 judgements that one hand is pushed forward to mistake draw garden for left hand, can from figure
Go out, more identical gesture, which easily produces, to be obscured, and causes to judge by accident.For this reason, when defining specific gesture, selective discrimination degree of trying one's best
Bigger gesture motion, so that system realizes gesture identification with higher precision.
Claims (1)
1. a kind of gesture identification method based on Wi-Fi signal, it is characterised in that specifically include following steps：
Step 1, transmitting terminal send signal, and user makes various gestures between transmitting terminal and receiving terminal, and signal is produced and is disturbed
Dynamic, two reception antennas of receiving terminal are respectively received signal G1 and G3 after being disturbed by gesture；
Step 2, pre-processes signal G1 and G3, and the pretreatment includes normalized, conjugation is handled at peaceful slide
Reason, obtains pretreated signal S1；
Step 3, hand signal GS to be identified is extracted according to signal S1；
Step 4, builds template hand signal storehouse, multiple template hand signal is included in template hand signal storehouse；By hand to be identified
Gesture signal GS is matched with multiple template hand signal respectively, obtains hand signal GS to be identified and multiple template hand signal
Between matching distance, wherein gesture and gesture to be identified represented by template hand signal corresponding to minimum matching distance
The gesture that signal represents is consistent；
The implementation of the step 2 includes：
Step 2.1:Signal G1 and G3 are normalized, respectively obtain signal G4 and G5 after normalization；
Step 2.2:Conjugation processing is carried out to the signal G4 and G5 after normalization, obtains removing the signal S after noise；
Step 2.3:Signal S after denoising is smoothed, obtains the signal S1 after smoothing processing；
The implementation of the step 3 includes：
Step 3.1：According to signal S1, sliding window amplitude and matrix A [n] are calculated, n represents data sample number：
<mrow>
<mi>A</mi>
<mo>&lsqb;</mo>
<mi>n</mi>
<mo>&rsqb;</mo>
<mo>=</mo>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
<mo>+</mo>
<mi>k</mi>
<mo>*</mo>
<mi>o</mi>
<mi>f</mi>
<mi>f</mi>
<mi>s</mi>
<mi>e</mi>
<mi>t</mi>
</mrow>
<mrow>
<mi>w</mi>
<mi>i</mi>
<mi>n</mi>
<mo>+</mo>
<mi>o</mi>
<mi>f</mi>
<mi>f</mi>
<mi>s</mi>
<mi>e</mi>
<mi>t</mi>
<mo>*</mo>
<mi>k</mi>
</mrow>
</munderover>
<mi>S</mi>
<msup>
<mn>1</mn>
<mn>2</mn>
</msup>
<mo>&lsqb;</mo>
<mi>i</mi>
<mo>&rsqb;</mo>
</mrow>
In above formula,Wherein, win is the size of sliding window, and offset is inclined every time for sliding window
Move incrementss；M is the sum of the data in signal S1；
Step 3.2：Determine gesture data constraints；
Step 3.3：Hand signal GS to be identified is obtained according to sliding window amplitude and matrix A [n] and gesture data constraints；
The implementation of the step 4 includes：
Step 4.1：Template hand signal storehouse is built, multiple template hand signal ref [k], k tables are included in template hand signal storehouse
Show kth kind template hand signal, single template hand signal data are expressed as ref (i), and 1≤i≤d, wherein d represent single mould
The number of the data point of plate gesture data；
Step 4.2：Hand signal GS to be identified is matched with multiple template hand signal respectively, obtains gesture letter to be identified
Matching distance number between GS and multiple template hand signal, concrete methods of realizing are as follows：
Hand signal data to be identified are GS (j), and 1≤j≤n, wherein n represent the number of the data point of gesture data to be identified；
If d=n, hand signal GS to be identified and all template hand signal ref are calculated by Euclidean distance formula
The matching distance value DS [k] of [k], k represent kth kind template gesture, the template hand signal institute corresponding to minimum matching distance
The gesture of expression is consistent with the gesture that hand signal to be identified represents；
If d ≠ n, the Distance matrix D (i, j) of d*n, 1≤i≤d, 1≤j≤n are built；Obtained according to Distance matrix D (i, j) and wait to know
The matching distance value DS [k] of other hand signal and all template hand signals, k represent kth kind template gesture, minimum matching away from
It is consistent with the gesture that hand signal to be identified represents from the gesture represented by corresponding template hand signal；
The implementation of the step 3.3 includes：
Step 3.3.1：One section of non-gesture state data is gathered, is averaged as threshold value threshold；
Step 3.3.2：The threshold value threshold extraction hand signals GS obtained according to step 3.3.1；
Step 3.3.2.1:With A [1] for starting point, A [1] represents first data of sliding window amplitude and matrix A, judges one by one
| A [n] | whether more than threshold value threshold, obtain in A [n] | A [n] | continuous the first segment data for being more than threshold value threshold
Section A [Tstart]~A [Tend], data interval A [Tstart]~A [Tend] number of samples be num；Judge num whether in section
In [Min, Max], wherein, Min is gesture data duration minimum value, and Max is gesture data duration maximum, if
It is to go to step 3.3.2.2, if not, going to step 3.3.2.3；
Step 3.3.2.2：Obtain A [Tstart]~A [Tend] data S1 [T in corresponding signal S1start* offset]~S1
[Tend* offset], offset deviates incrementss every time for sliding window, calculates S1 [Tatart* offset]~S1 [Tend*
Offset] in the sum of all data absolute values V；If V is more than minimum threshold min_gesture, then data point S1 is thought
[Tstart* offset]~S1 [Tend* offset] it is gesture data GS, otherwise data point S1 [Tstart* offset]~S1 [Tend*
Offset] it is non-gesture data, with A [Tend+ 1] it is starting point, goes to step 3.3.2.1；
Step 3.3.2.3：When num values are more than Max, data point S1 [Tstart* offset]~S1 [Tend* offset] it is non-hand
Gesture data, with A [Tend+ 1] it is starting point, goes to step 3.3.2.1；When num values are less than Min, continue to A [Tend] after data
Judge one by one | A [n] | whether more than threshold value threshold, obtain the first segment data interval for being continuously less than threshold value threshold
A[Tend+ 1]~A [Tend1], statistics section A [Tend+ 1]~A [Tend1] data amount check be num1；Judge whether num1 is big
In Min, if so, 3.3.2.4 is gone to step, if not, going to step 3.3.2.5；
Step 3.3.2.4：Data interval A [T may determine that according to gesture duration featurestart]~A [Tend1] it is non-gesture
Data, with A [Tend1+ 1] it is starting point, goes to step 3.3.2.1；
Step 3.3.2.5：Continue to A [Tend1] after data judge one by one | A [n] | whether more than threshold value threshold, obtain
To the continuous second segment data interval A [T for being more than threshold value thresholdend1+ 1]~A [Tend2], statistics section A [Tend1+
1]~A [Tend2] data amount check be num2；The sum of num, num1 and num2 are judged whether in section [Min, Max], if
The sum of three numbers then judge data interval A [T in the sectionstart]~A [Tend2] corresponding to signal S1 in data S1
[Tstart* offset]~S1 [Tend2* offset] it is gesture data GS；Otherwise it is non-gesture data, with A [Tend2+ 1] it is
Point, goes to step 3.3.2.1.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510939043.0A CN105573498B (en) | 2015-12-15 | 2015-12-15 | A kind of gesture identification method based on Wi-Fi signal |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510939043.0A CN105573498B (en) | 2015-12-15 | 2015-12-15 | A kind of gesture identification method based on Wi-Fi signal |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105573498A CN105573498A (en) | 2016-05-11 |
| CN105573498B true CN105573498B (en) | 2018-05-08 |
Family
ID=55883728
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510939043.0A Active CN105573498B (en) | 2015-12-15 | 2015-12-15 | A kind of gesture identification method based on Wi-Fi signal |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105573498B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US11995242B2 (en) | 2020-07-15 | 2024-05-28 | Google Llc | Detecting contactless gestures using radio frequency |
Families Citing this family (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106066995B (en) * | 2016-05-25 | 2019-10-11 | 西安交通大学 | A kind of wireless unbundling human body behavioral value algorithm |
| CN106774894A (en) * | 2016-12-16 | 2017-05-31 | 重庆大学 | Interactive teaching methods and interactive system based on gesture |
| CN107067031B (en) * | 2017-03-29 | 2020-10-23 | 西北大学 | Calligraphy posture automatic identification method based on Wi-Fi signal |
| CN107102729A (en) * | 2017-04-05 | 2017-08-29 | 河南师范大学 | A kind of PPT Demonstration Control Systems based on CSI gesture identifications |
| CN107241698B (en) * | 2017-07-17 | 2020-03-24 | 北京大学 | Non-contact perception tracking method |
| CN108459706A (en) * | 2018-01-24 | 2018-08-28 | 重庆邮电大学 | Wi-Fi gesture identification methods based on relative movement orbit tracking |
| CN109407833A (en) * | 2018-09-30 | 2019-03-01 | Oppo广东移动通信有限公司 | Manipulate method, apparatus, electronic equipment and the storage medium of electronic equipment |
| CN109460716A (en) * | 2018-10-19 | 2019-03-12 | 大连理工大学 | A kind of sign language wireless-identification device and method |
| CN110110580B (en) * | 2019-03-12 | 2023-04-07 | 西北大学 | Wi-Fi signal-oriented sign language isolated word recognition network construction and classification method |
| US11442550B2 (en) * | 2019-05-06 | 2022-09-13 | Samsung Electronics Co., Ltd. | Methods for gesture recognition and control |
| CN110458118B (en) * | 2019-08-14 | 2022-08-12 | 南京邮电大学 | Simple sign language identification method based on channel state information |
| EP3966668A1 (en) * | 2020-07-15 | 2022-03-16 | Google LLC | Detecting contactless gestures using radio frequency |
| CN112218303B (en) * | 2020-09-28 | 2022-02-18 | 上海交通大学 | Signal conversion method based on Wi-Fi identification system |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103971108A (en) * | 2014-05-28 | 2014-08-06 | 北京邮电大学 | Wireless communication-based human body posture recognition method and device |
| CN104503575A (en) * | 2014-12-18 | 2015-04-08 | 大连理工大学 | Method for designing low-power-consumption gesture recognition circuit device |
-
2015
- 2015-12-15 CN CN201510939043.0A patent/CN105573498B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103971108A (en) * | 2014-05-28 | 2014-08-06 | 北京邮电大学 | Wireless communication-based human body posture recognition method and device |
| CN104503575A (en) * | 2014-12-18 | 2015-04-08 | 大连理工大学 | Method for designing low-power-consumption gesture recognition circuit device |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US11995242B2 (en) | 2020-07-15 | 2024-05-28 | Google Llc | Detecting contactless gestures using radio frequency |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105573498A (en) | 2016-05-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105573498B (en) | A kind of gesture identification method based on Wi-Fi signal | |
| Yang et al. | MU-ID: Multi-user identification through gaits using millimeter wave radios | |
| US10061389B2 (en) | Gesture recognition system and gesture recognition method | |
| CN111399642B (en) | Gesture recognition method and device, mobile terminal and storage medium | |
| CN109325967A (en) | Method for tracking target, device, medium and equipment | |
| CN104700069B (en) | A kind of unbundling formula radio-frequency (RF) tag body-building action recognition monitoring system and method thereof | |
| EP3497467A1 (en) | Control system and control processing method and apparatus | |
| CN108171278B (en) | Motion pattern recognition method and system based on motion training data | |
| CN110113116B (en) | Human behavior identification method based on WIFI channel information | |
| CN103839040A (en) | Gesture identification method and device based on depth images | |
| CN105844216A (en) | Detection and matching mechanism for recognition of handwritten letters using WiFi signals | |
| CN107992792A (en) | A kind of aerial handwritten Chinese character recognition system and method based on acceleration transducer | |
| Chen et al. | Human behavior recognition using Wi-Fi CSI: Challenges and opportunities | |
| CN113609976B (en) | Direction-sensitive multi-gesture recognition system and method based on WiFi equipment | |
| CN106685590A (en) | Channel state information and KNN (K-Nearest Neighbor)-based indoor human body orientation recognition method | |
| CN104834412B (en) | A kind of touch terminal based on contactless gesture identification | |
| CN103530892A (en) | Kinect sensor based two-hand tracking method and device | |
| CN110059612A (en) | A kind of gesture identification method and system that the position based on channel state information is unrelated | |
| CN103105924A (en) | Man-machine interaction method and device | |
| CN110866468A (en) | Gesture recognition system and method based on passive RFID | |
| CN113918019A (en) | Gesture recognition control method and device for terminal equipment, terminal equipment and medium | |
| Uysal et al. | RF-Wri: An efficient framework for RF-based device-free air-writing recognition | |
| CN106599873A (en) | Figure identity identification method based on three-dimensional attitude information | |
| Pan et al. | Dynamic hand gesture detection and recognition with WiFi signal based on 1d-CNN | |
| Zhu et al. | Wi-ATCN: Attentional temporal convolutional network for human action prediction using WiFi channel state information |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CB03 | Change of inventor or designer information |
Inventor after: Chen Xiaojiang Inventor after: Wang Anwen Inventor after: Ren Yuhui Inventor after: Guo Songtao Inventor after: He Gang Inventor after: Xu Dan Inventor after: Liu Dongdong Inventor after: Chen Feng Inventor after: Wang Liang Inventor after: Li Wei Inventor after: Tang Zhanyong Inventor after: Peng Yao Inventor after: Zhang Jie Inventor before: Liu Dongdong Inventor before: Guo Songtao Inventor before: He Gang Inventor before: Wang Liang Inventor before: Li Wei Inventor before: Chen Xiaojiang Inventor before: Tang Zhanyong Inventor before: Peng Yao Inventor before: Zhang Jie Inventor before: Wang Anwen Inventor before: Ren Yuhui |
|
| CB03 | Change of inventor or designer information |