CN105373215B - Dynamic radio gesture identification method with decoding is encoded based on gesture - Google Patents
Dynamic radio gesture identification method with decoding is encoded based on gesture Download PDFInfo
- Publication number
- CN105373215B CN105373215B CN201410423086.9A CN201410423086A CN105373215B CN 105373215 B CN105373215 B CN 105373215B CN 201410423086 A CN201410423086 A CN 201410423086A CN 105373215 B CN105373215 B CN 105373215B
- Authority
- CN
- China
- Prior art keywords
- gesture
- basic
- coding
- signal
- sub
- 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
Landscapes
- User Interface Of Digital Computer (AREA)
Abstract
The present invention discloses a kind of dynamic radio gesture identification method encoded based on gesture with decoding, comprises the following steps：Construct basic gesture collection, basic gesture off-line training, gesture coding, order hand signal acquisition to be checked, the identification of order hand signal, hand signal decoding.The dynamic radio gesture identification method that the present invention uses, suitable for a variety of radio and ultrasonic wave gesture recognition systems characterized by hand signal amplitude, phase, direction of arrival or Doppler frequency shift, versatility is good, and high to the identification probability of wireless gesture, good reliability.
Description
Technical field
The invention belongs to technical field of hand gesture recognition in man-machine interaction, particularly a kind of identification probability is high, good reliability
Dynamic radio gesture identification method with decoding is encoded based on gesture.
Background technology
In the various necks such as intelligent robot, intelligent television, smart mobile phone, computer, game machine, intelligent automobile, smart home
Domain, computer is wirelessly operated using dynamic gesture turns into new man-machine interaction mode.
Existing dynamic radio gesture identification mode includes wearable and naked hand gesture recognition system, as document " B.Raj,
K.Kalgaonkar,C.Harrison,and P.Dietz,Ultrasonic Doppler sensing in HCI,IEEE
Pervasive Computing, vol.11, no.2, pp.24-29, Feb.2012 " are described.Wearable system, which need to dress to include, to be added
The gloves of the sensors such as speedometer, magnetometer identify gesture, in-convenience in use.Naked hand gesture recognition system is connect using non-physical
The mode of touching interacts with machine, is not required to dress special gloves to identify gesture, easy to use.Most common realization rate is vision
(light wave), ultrasonic wave, radio wave identifying system.
Visual gesture service system is using camera shooting image and identifies gesture, but following limitation be present：1st, identification away from
From near, the distance increase image effect variation that then camera is shot；2nd, system needs continuous light, in the dark thrashing；
3rd, sensitive, typically inapplicable movement or outdoor environment are changed to light；4th, line-of-sight propagation is needed between people-machine, camera and the human world are not
There can be obstacle；5th, camera may invade privacy of user.
Radio, which involves the naked hand Gesture Recognition of ultrasonic wave, has many advantages：
(1) is easy to use without wearable device.
(2) can realize more remote gesture control.
(3) is small by such environmental effects such as temperature, brightness.It can be used in dark, outdoor environment.
(4) can work in nlos environment.
(5) does not influence the privacy of user.
With specific reference to " S.Gupta, D.Morris, S.Patel, and D.Tan, Soundwave:using the
doppler effect to sense gestures,Proceedings of the SIGCHI Conference on
Human Factors in Computing Systems,May 05-10,2012,Austin,Texas,USA,2012”、
“F.Adib and D.Katabi,See through walls with Wi-Fi！Proceedings of the ACM
SIGCOMM 2013 conference on SIGCOMM,August 12-16,2013,Hong Kong,China,2013”、
“Q.Pu,S.Gupta,S.Gollakota,and S.Patel,Whole-home gesture recognition using
wireless signals,ACM International Conference on Mobile Computing and
Networking(MOBICOM),2013”、“F.Adib,Z.Kabelac,D.Katabi,and R.C.Miller,3D
Tracking via body radio reflections, Usenix NSDI, 2014 " and " B.Kellogg, V.Talla, and
S.Gollakota,Bringing gesture recognition to all devices,Usenix NSDI,2014”。
But in the wireless channel that radio involves residing for ultrasonic wave, realize that the correct identification of gesture faces a series of choose
War, including：
(1) channel circumstances influence：Many unfavorable factors such as interference, decline, multipath in wireless channel be present, add nothing
The probability of misrecognition of line gesture recognition system.
(2) gesture designs influence：Current gesture design is in the state of a kind of " one is unified approach ", lacks systematicness
Reliable Design method.Design the big gesture set of discrimination so that also can in set even if two gestures apart from minimum
Distinguish, it is generally relatively difficult.This is the immanent cause of gesture misrecognition.
(3) manipulators influence：Gesture is completed by people, the possible maloperation of people, so as to cause the gesture command of mistake.
On the other hand, in numerous applications, the reliable Gesture Recognition demand low to probability of misrecognition is urgent, such as military
Or the control of bomb disposal robot, military aircraft landing commander, intelligent automobile control etc. in anti-terrorism.Therefore, high reliability gesture is known
Contradiction be present between other active demand and gesture identification result are unreliable.
In a word, the problem of prior art is present be：Identification probability is not high enough, it is impossible to meets the need of high reliability gesture control
Ask, versatility is not strong.
The content of the invention
It is an object of the invention to provide a kind of dynamic radio gesture identification method for being encoded and being decoded based on gesture, identification
Probability is high, and good reliability is versatile.
The technical solution for realizing the object of the invention is：A kind of encoded based on gesture is known with the dynamic radio gesture decoded
Other method, comprises the following steps：
10) basic gesture collection is constructed：According to multiple basic gestures, basic gesture collection is constructed；
20) basic gesture off-line training：Each basic gesture, the transmitting that repeated detection reflects through it are concentrated to basic gesture
The signal that machine is launched, the feature of each basic gesture is drawn, store the feature of all basic gestures, establish basic gesture feature
Storehouse；
30) gesture encodes：Basic gesture is encoded into the order gesture corresponding with gesture control order, constructs the life
Gesture is made to include basic gesture assembly coding, error detection coding and Error Correction of Coding；
40) order hand signal to be checked obtains：Emitter transmission signal, receiver receive the letter reflected through order gesture
Number, obtain order hand signal to be checked；
50) order hand signal identifies：Order hand signal to be checked is divided into more sub- gestures, every sub- gesture corresponding one
Individual unknown basic gesture, realize that order hand signal identifies by each sub- gesture identification, each sub- gesture identification is detection
The signal characteristic of each sub- gesture, each sub- gesture feature is contrasted with basic gesture feature successively, found out and each sub- gesture phase
The basic gesture of matching, these basic gestures combine successively, form order hand signal court verdict to be checked；
60) hand signal decodes：Error detection decoding and error-correcting decoding are carried out to hand signal court verdict to be checked, obtains and treats
Examine gesture control order corresponding to hand signal.
The present invention compared with prior art, its remarkable advantage：Identification probability is high, and good reliability is versatile.
Above-mentioned advantage reaches by the following method：
1st, gesture species is simplified：Only use a few basic gesture, and make full use of communication in coding theory, you can construct
Other gestures, avoid gesture amount it is big when cumbersome gesture training process；When the gesture number in basic gesture set is few, hold
Easily construction basic gesture makes gesture collection minimum range (distance i.e. between two minimum gestures of distance) larger；It is and minimum between gesture
Distance is bigger, and probability of miscarriage of justice is smaller；
2nd, versatility is improved：Gesture assembly coding and Jian Cuo ﹑ Error Corrections of Coding independently of specific system (such as ultrasonic wave,
Radio, Doppler or signal amplitude), there is universality；
3rd, accidental basic gesture erroneous judgement can be corrected：After gesture digitlization, error correction coding in digital communication can be used,
When wireless channel decline, the influence of interference or artificial maloperation, when causing erroneous judgement once in a while occur in basic gesture judgement, can entangle
Positive erroneous judgement, ensures correct gesture identification.
The present invention is described in further detail with reference to the accompanying drawings and detailed description.
Brief description of the drawings
Fig. 1 is the dynamic radio gesture identification method flow chart that the present invention is encoded and decoded based on gesture.
Fig. 2 a are ternary basic gesture schematic diagram (gesture 0).
Fig. 2 b are ternary basic gesture schematic diagram (gesture 1).
Fig. 2 c are ternary basic gesture schematic diagram (gesture 2).
Fig. 3 a are a kind of assembly coding schematic diagram of certain gestures under ternary fixed-length coding pattern.
Fig. 3 b are a kind of error detection coding schematic diagram of certain gestures under ternary fixed-length coding pattern.
Fig. 3 c are a kind of Error Correction of Coding schematic diagram of certain gestures under ternary fixed-length coding pattern.
Fig. 4 a are the coding schedule of the non-isometric coding of ternary of nine kinds of gesture commands of identification.
Fig. 4 b are that the error detection of the non-isometric coding of ternary of nine kinds of gesture commands of identification decodes schematic diagram.
Fig. 5 a are the coding schedule of the non-isometric coding of ternary of nine kinds of gesture commands of identification.
Fig. 5 b are the error-correcting decoding schematic diagram of the non-isometric coding of ternary of nine kinds of gesture commands of identification.
Embodiment
As shown in figure 1, the present invention encodes the dynamic radio gesture identification method with decoding, including following step based on gesture
Suddenly：
10) basic gesture collection is constructed：According to multiple basic gestures, basic gesture collection is constructed；
The multiple basic gesture meets：Each basic gesture is divided into etc. to multiple run duration sections of quantity, same
In the period of one sequence number, change between each basic gesture relative to the direction of motion of receiver, motion amplitude or motion model
Enclose, make that minimum norm distance is maximum between the characteristic vector of each basic gesture or makes the characteristic vector of each basic gesture orthogonal
(even if vector angle is 90 degree).
Wherein, two vector a=[a1,a2,…aN] ﹑ b=[b1,b2,…bN] angle calcu-lation formula：
Norm distance calculation formula between two vector a ﹑ b：
Make each a kind of maximum method of basic gesture characteristic vector norm distance：Basic gesture number N is determined, designs one group
M(M>N) individual alternative basic gesture, measure each alternative gesture feature vector respectively once, therefrom choose a N rank subset, make this
The distance of two minimum basic gestures of characteristic vector norm distance is maximum in subset.
Make a kind of orthogonal method of characteristic vector of each basic gesture：If basic gesture number is three kinds, and basic gesture
It is divided into 12 sections, then constructs three kinds of basic gestures and cause its characteristic vector a1、a2、a3There is following form respectively after taking symbol：
a1=[1 1111 1-1-1-1-1-1-1]
a2=[1 1111111111 1]
a3=[- 1-1-1 11111 1-1-1-1]
It can verify that, above-mentioned three vectors pairwise orthogonal.The orthogonal vector of other forms also includes a form and its various changes
Shape：
a1=[1 0000000000 0]
a2=[0 0001000000 0]
a3=[0 0000000100 0]
And
a1=[1 0000100001 0]
a2=[0 0011000010 0]
a3=[0 0100001100 0]
The present invention will not enumerate.
In embodiment, obtained using characteristic vector norm apart from maximum solution as shown in Figure 2 a kind of ternary basic
Gesture schematic diagram.
In embodiment, three kinds of basic gestures are selected, basic gesture is mapped as " 0 ", " 1 ", " 2 ", encoded using ternary.Will
Feature of the Doppler frequency shift values as gesture caused by gesture motion.Doppler frequency displacements and the relative velocity and direction phase of motion
Close.Wherein, gesture " 0 "：Circular motion is done, away from receiver after first being moved towards receiver, returns to starting point.Gesture " 1 "：Do circle
Arc moves, and first away from being moved after receiver towards receiver, returns to starting point.Gesture " 2 "：Straight line fortune is done on man-machine line direction
It is dynamic, first away from being moved after receiver towards receiver, starting point is returned to, moves, has returned to away from receiver after being re-directed towards receiver
Point.
20) basic gesture off-line training：Each basic gesture, the transmitting that repeated detection reflects through it are concentrated to basic gesture
The signal that machine is launched, the feature of each basic gesture is drawn, store the feature of all basic gestures, establish basic gesture feature
Storehouse；
Basic gesture off-line training (20) step includes：
21) basic gesture is segmented：To each basic gesture, its complete procedure is divided into etc. to the multistage of duration；
In described basic gesture segmentation (21) step, basic gesture division number is 5~30 sections.
22) characteristic scalar obtains：By signal characteristic discretization corresponding to every section of gesture, the signal characteristic of each section of gesture is obtained
Scalar；
The characteristic scalar obtains (22) step：
The multiple measured value of each segment signal is averaged or the spy using the median of multiple measured value as each segment signal
Scalar is levied, each segment signal characteristic scalar uses the absolute figure of measured value or only takes its symbol.
The characteristic scalar is obtained in (22) step, and the signal characteristic includes signal amplitude, phase, signal direction of arrival
Or Doppler frequency shift is with the variable quantity of gesture motion.
23) characteristic vector obtains：The signal characteristic scalar of whole basic gesture is combined successively, forms the basic gesture
Character vector of signals；
24) basic gesture feature database is established：The character vector of signals of each basic gesture forms basic gesture feature database.
In embodiment, basic gesture is divided into 12 sections, measures feature of each section of the Doppler frequency shift values as gesture, will
Multiple measured value obtains each section of Doppler frequency displacements after being averaged, when completing same gesture due to different user, its characteristic value arrow
Amount there may be difference, in order to eliminate difference, takes the symbol of Doppler frequency displacements and combines the Doppler frequencies for forming each gesture
Vector is moved, is designated as G respectively0、G1、G2.According to Doppler frequency shift properties, each basic gesture of training stage measurement
A kind of form of Doppler frequency displacement vectors is as follows：G0=[1 1111 1-1-1-1-1-1-1], G1=[- 1-1-1
- 1-1-1 11111 1], G2=[1 1 1-1-1-1-1-1-1 11 1].Above-mentioned three vector forms basic
Gesture feature storehouse.
30) gesture encodes：Basic gesture is encoded into the order gesture corresponding with gesture control order, constructs the life
Gesture is made to include basic gesture assembly coding, error detection coding and Error Correction of Coding；
Described gesture coding (30) step includes：
31) coding mode judges：Judge gesture coding mode, encoded if non-isometric degree, then go to fault-tolerant encoding pattern
Judge (33) step；
32) gesture generation is combined：Combined using the basic gesture of identical quantity, form combination gesture, realize gesture command
To the mapping of combination gesture.
33) fault-tolerant encoding mode decision：Judge fault-tolerant encoding pattern, if Error Correction of Coding pattern, then go to Error Correction of Coding
Generate (35) step；
34) error detection coding generates：Error detection coding is carried out to gesture command or combination gesture, according to equal length code, then
It is different to combine any two code character at least two code elements in the coding schedule of gesture error detection coding, is encoded according to non-isometric degree,
Then either length is different or at least two code elements are different for any two code character in the coding schedule of gesture command error detection coding；
35) Error Correction of Coding generates：Error Correction of Coding is carried out to gesture command or combination gesture, according to equal length code, then
Any two code character at least three symbol differences in the coding schedule of gesture Error Correction of Coding are combined, are encoded according to non-isometric degree,
Then either length is different or at least three symbol differences for any two code character in the coding schedule of gesture command Error Correction of Coding.
As shown in figure 3, it is a kind of assembly coding of certain gestures, error detection coding and error correction under ternary fixed-length coding pattern
Coding schematic diagram.
Isometric assembly coding scheme：In the embodiment of ternary fixed-length coding pattern, three kinds of basic gestures can group shown in Fig. 2
Nine kinds of gestures are synthesized, nine kinds of gestures be can be used to control nine kinds of orders, and nine kinds of gesture one kind may be encoded to
Isometric error detection coding scheme：Further doing combination gesture error detection coding to combinations thereof coding, (misoperation probability is not
Gao Shi, operator scheme are set as error detection coding, and gesture simply avoids maloperation simultaneously).The code length of error detection coding is 3,
Increase an error detection bit i.e. on the basis of assembly coding.A kind of error detection coding method of combinations thereof coding is as follows：
(1) by above-mentioned nine kinds combination gestures by being divided longitudinally into three groups, first group is 00 01 02.First group of error detection coding
3rd is combined for 0,1,2 not repeated arrangement, totally 6 kinds：000 011 022、000 012 021、001 010 022、001
012 020、002 010 021、002 011 020；
(2) second and third group the 3rd is respectively the cyclic shift of upper one group the 3rd, and cyclic shift includes moving to left and moving to right
Two kinds, and second and third group of direction of displacement is consistent, realizes for totally two kinds, altogether 12 kinds of coded systems.
(3) above-mentioned coded system minimum range is 2：Ensure that any two code character at least two code elements are different, occur one
When individual symbol erroneous judgement is determined, it can be found that this mistake.Such as under the first coded system, when " 000 " is mistaken for " 001 ", it can find
This decoding error.
According to methods described, all 12 kinds of error detection codings are：
On the basis of above-mentioned gesture error detection coding, continue one coding of increase and gesture Error Correction of Coding can be achieved, be i.e. error correction is compiled
Code length is 4, and one group of Error Correction of Coding removes one group of error detection coding of the 4th remaining three composition.The embodiment of the present invention is with the first
Exemplified by Error Correction of Coding, illustrate Error Correction of Coding construction process.With other the 11 kinds methods for constructing Error Correction of Coding of above-mentioned error detection coding
It is identical.
Isometric error correction coding scheme：Using ternary (4,2) code, i.e. code length is 4, wherein 2 bit information positions.It is each
Individual error detection coding is by longitudinal directionDivide three groups,First groupFourPosition is combined for 0,1,2 not repeated arrangement, totally 6 kinds；Second and third group theFourPosition is respectively each upper one group theFour(i.e. second group the of the cyclic shift of positionFourPosition is first group theFourThe cyclic shift of position,
3rd groupFourPosition is second group theFourThe cyclic shift of position), second and third group4thDirection of displacement is consistent, and circulate and move Position mode is just opposite with the cyclic shift mode of the 3rd.Each group of error detection coding finally derives 6 groups of Error Corrections of Coding, altogether
72 kinds of Error Correction of Coding modes.Error Correction of Coding minimum range is 3：Any two encodes at least 3 symbol differences in one group of coding.
When error detection coding uses in 12 groups first group：
000 011 022
101 112 120
202 210 221
Corresponding 6 groups of Error Corrections of Coding are：
In the embodiment of the present invention, for the ease of memory, using second of Error Correction of Coding
0000 0112 0221
1011 1120 1202
2022 2101 2210
The characteristics of this group of Error Correction of Coding：The 1st each encoded+the 2=the 3rd, the 2nd+the 3=the 4th (this
Place adds up to the addition of mould 3).Therefore, command number corresponding to gesture control order need to simply only be remembered, its coded system can calculate
Arrive.In being controlled such as robot, " advance " corresponding No. 2 orders, then combinational code is 01, error detection coding 011, and Error Correction of Coding is
0112。
Non-isometric error detection coding scheme：Present invention simultaneously provides a kind of non-isometric gesture command of ternaryError detectionCoding, still
By taking nine kinds of gesture commands as an example, form is as follows after its error detection coding：
Present invention simultaneously provides a kind of binary system is non-isometricError detectionCoding, basic gesture are mapped as " 0 ", " 1 " two codings
Symbol.The corresponding gesture of " 0 ", " 1 " is same as above.Still by taking nine kinds of gesture commands as an example, form is as follows after its error detection coding：
Non-isometric error correction coding scheme：Present invention simultaneously provides a kind of non-isometric gesture command of ternaryError correctionCoding, still
By taking nine kinds of gesture commands as an example, form is as follows after its Error Correction of Coding：
Present invention simultaneously provides a kind of binary system is non-isometricError correctionCoding, basic gesture are mapped as " 0 ", " 1 " two codings
Symbol.The corresponding gesture of " 0 ", " 1 " is same as above.Still by taking nine kinds of gesture commands as an example, form after its assembly coding and Error Correction of Coding
It is as follows：
In above-mentioned binary system and ternary error detection coding, in addition to a unique primitive encoding " 0 ", all code lengths are not small
In 2 any different to coding or code length, or at least 2 symbol differences, thus there is error detecing capability.
Equally, above-mentioned binary system and ternary Error Correction of Coding each be different from it on code length or code symbols
He.As ternary is non-isometricError correctionIn coding, the unique primitive encoding " 0 " of code length and unique binary coding " 11 ", its
Remaining length is identical and code length be not less than 3 coding at least three symbol differences, such as " 000 " and " 111 ", thus above-mentioned volume
Two, the error correcting capability source aspect of code：(1) coding of the length not less than 3 is determined by Error Correction of Coding to correct gesture erroneous judgement；(2)
Coding of the length less than 3 avoids gesture from judging by accident by unique mapping between gesture and code length.Meanwhile shorter coding point
The more frequently used control command of dispensing.
By above-mentioned thinking, slightly change the binary system that can also construct other forms and ternary coding, the present invention is not another
One enumerates.
40) order hand signal to be checked obtains：Emitter transmission signal, receiver receive the letter reflected through order gesture
Number, obtain order hand signal to be checked；
The hand signal to be checked is obtained in (40) step, and transmission signal includes：Tone signal, two-tone signal, orthogonal frequency
Division multiplex signal, frequency modulation continuous wave signal.
50) order hand signal identifies：Order hand signal to be checked is divided into more sub- gestures, every sub- gesture corresponding one
Individual unknown basic gesture, realize that order hand signal identifies by each sub- gesture identification, each sub- gesture identification is detection
The signal characteristic of each sub- gesture, each sub- gesture feature is contrasted with basic gesture feature successively, found out and each sub- gesture phase
The basic gesture of matching, these basic gestures combine successively, form order hand signal court verdict to be checked；
In hand signal identification (50) step, the sub- gesture identification includes：
51) sub- hand signal segmentation：To each sub- hand signal, its complete procedure is divided into etc. to the multistage of duration, its
Division number is segmented in (21) step that division number is identical with the basic gesture, calculates the signal characteristic scalar of each segmentation,
Sub- hand signal characteristic vector is sequentially combined into, and according to pattern is preset, retains sub- hand signal characteristic vector
In each scalar absolute figure or only retain its symbol；
52) signal characteristic type judges：Judge the signal characteristic type of sub- hand signal, be such as the symbol of measured value,
Then go to norm calculation (54) step；
53) angle calcu-lation：Calculate the folder of sub- hand signal characteristic vector and all characteristic vectors in basic gesture feature database
Angle, go to gesture matching (55) step；
54) norm calculation：Calculate the difference of all characteristic vectors in sub- hand signal characteristic vector and basic gesture feature database
Norm；
55) sub- gesture matching：Find in basic gesture feature database with the angle of sub- hand signal characteristic vector or with sub- hand
The basic gesture of the Norm minimum of the difference of gesture character vector of signals, as the basic gesture to match with sub- hand signal.
The division of measurement process neutron gesture needs to distinguish the beginning and end of gesture, and a kind of existing method is in different gestures
Between stop the of short duration time, to identify that new gesture starts, with specific reference to " Q.Pu, S.Gupta, S.Gollakota, and
S.Patel,Whole-home gesture recognition using wireless signals,ACM
International Conference on Mobile Computing and Networking(MOBICOM),2013”.Son
In gesture identification embodiment, to every sub- gesture, sub- hand signal is divided into 12 sections, measures the Doppler frequency displacements of each segment signal
Value, the symbol of measured value is taken to form the characteristic vector of measurement.Calculate the characteristic vector of measurement respectively and the training stage obtain it is each
The Doppler frequency displacement vectors G of individual basic gesture0、G1、G2The distance between, Doppler frequency displacements vector and the characteristic vector of measurement
The minimum basic gesture of distance is the gesture adjudicated.Herein, the distance between two vectors are defined as the norm of the difference of two vectors.
Such as when basic gesture is gesture " 0 ", measurement error causes the Doppler frequency displacements vector of measurement to be Gm=[- 11
1 1 1 1 -1 1 -1 -1 -1 -1].Then GmWith G0、G1、G2The distance between d0,m、d1,m、d2,mRespectivelyObviously, the gesture for adjudicating measurement is basic gesture " 0 ".
60) hand signal decodes：Error detection decoding and error-correcting decoding are carried out to hand signal court verdict to be checked, obtains and treats
Examine gesture control order corresponding to hand signal.
Described hand signal decoding (60) step includes：
61) fault-tolerant encoding mode decision：Judge gesture fault-tolerant encoding pattern, if error correction mode, then go to gesture error correction
Decode (63) step；
62) gesture error detection decodes：Whether judgement gesture identification there is mistake, stops decoding, gesture control if mistake is found
Invalid operation is made for, otherwise decodes and goes to gesture control order output (64) step；
63) gesture error-correcting decoding：Using optimal maximum-likelihood decoding or minimal distance docoding, correct and go out in gesture identification
Existing mistake, obtain decoding result；
64) gesture control order exports：Gesture control order corresponding with decoding result is exported.
After the completion of basic gesture judgement, the measured value of the i.e. original gesture coding of one group of coding is obtained, the present invention is referred to as " measurement
Code ".If measurement error is excessive so that judging by accident once in a while occurs in above-mentioned basic gesture judgement, and " measurement code " will differ from original volume
Code.Further it can be handled by the fault-tolerant decoding of gesture：The gesture that the embodiment of the present invention gives nine kinds of gesture commands of identification is known
A kind of isometric error detection decoding of ternary of other system and error-correcting decoding, and the non-isometric error detection decoding of ternary and error-correcting decoding are translated
Code method.
The isometric error detection decoding of ternary：It is as shown in the table according to the coding characteristic of the isometric error detection coding table of ternary, judge
Whether the digits of " measurement code " meet following relation (wherein addition is the addition of mould 3)：1st+the 2=the 3rdSuch as phase
Deng, peel off the 3rd obtain decode result.It is such as unequal, then miscue is provided, current gesture operation is invalid.
000 011 022
101 112 120
202 210 221
The isometric error-correcting decoding of ternary：The distance of all codings in " measurement code " and following code table is calculated successively.Distance is fixed
Justice is：The different number of symbol is corresponded in two codings.As actual gesture is encoded to " 0000 ", measurement error causes judgement
" measurement code " is " 0002 ", " measurement code " and code table
0000 0112 0221
1011 1120 1202
2022 2101 2210
In it is each coding distance be：
1 2 3
3 4 2
2 3 4
Then " 0000 " and " 0002 " distance are minimum, after error-correcting decoding obtains " 0000 ", peel off third and fourth and are translated
Code result.
The non-isometric error detection decoding of ternary：Judge the length of " measurement code ", when code length is 1, directly output decoding
As a result " 0 ", when code length is 2, the distance of each binary coding in " measurement code " and code table is calculated, with reference to following code table,
It is [00 11 22] herein.If minimum range is zero, directly " measurement code " is exported as decoding；If minimum range is more than zero,
Miscue is provided, basic gesture judgement is wrong.When code length is 3, calculates each ternary in " measurement code " and code table and compile
The distance of code [000 011 022 101 202].If minimum range is zero, directly " measurement code " is exported as decoding；If most
Small distance is more than zero, provides miscue, and basic gesture judgement is wrong.
0 00 11
22 000 011
022 101 202
The non-isometric error-correcting decoding of ternary：Judge the length of " measurement code ", when code length is 1, directly output decoding
As a result " 0 ".When code length is 2, directly output decodes result " 11 ".When code length is 3, " measurement code " and code are calculated
The distance of each three primitive encoding in table, it is [000 111 222] herein with reference to following code table, with " measurement code " distance minimum
Three primitive encodings export as decoding.When code length be 4 when, processing with code length for 3 when it is identical.
0 11 000
111 222 0000
0111 0222 1012
According to decoding result, gesture control command can be obtained, and perform the corresponding command.
In the inventive method, only use a few basic gesture, and make full use of communicate in coding theory, you can construct it
His gesture, avoid gesture amount it is big when cumbersome gesture training process；When the gesture number in basic gesture set is few, easily
Construction basic gesture makes its minimum range (distance i.e. between two minimum gestures of distance) larger；And minimum range is got between gesture
Greatly, probability of miscarriage of justice is smaller；
In the inventive method, gesture assembly coding and Jian Cuo ﹑ Error Corrections of Coding independently of specific system (such as ultrasonic wave,
Radio, Doppler or signal amplitude), there is universality；
Gesture digitlization after, can be used digital communication in error correction coding, when wireless channel decline, interference influence or
Artificial maloperation, when causing erroneous judgement once in a while occur in basic gesture judgement, erroneous judgement can be corrected, ensures correct gesture identification.
So as to the dynamic radio gesture identification method for encoding and decoding based on gesture using the present invention, to wireless gesture
Identification probability is high, good reliability, and can be used for a variety of wireless gesture identifying systems.
Claims (10)
1. a kind of dynamic radio gesture identification method encoded based on gesture with decoding, it is characterised in that comprise the following steps：
10) basic gesture collection is constructed：According to multiple basic gestures, basic gesture collection is constructed；
20) basic gesture off-line training：Each basic gesture, the emitter institute that repeated detection reflects through it are concentrated to basic gesture
The signal of transmitting, the feature of each basic gesture is drawn, store the feature of all basic gestures, establish basic gesture feature database；
30) gesture encodes：Basic gesture is encoded into the order gesture corresponding with gesture control order, constructs the order hand
Gesture includes basic gesture assembly coding, error detection coding and Error Correction of Coding；
40) order hand signal to be checked obtains：Emitter transmission signal, receiver receive the signal reflected through order gesture, obtained
To order hand signal to be checked；
50) order hand signal identifies：Order hand signal to be checked is divided into more sub- gestures, corresponding one of every sub- gesture is not
The basic gesture known, realize that order hand signal identifies by each sub- gesture identification, each sub- gesture identification is each to detect
The signal characteristic of sub- gesture, each sub- gesture feature is contrasted with basic gesture feature successively, find out and match with each sub- gesture
Basic gesture, these basic gestures combine successively, form order hand signal court verdict to be checked；
60) hand signal decodes：Error detection decoding and error-correcting decoding are carried out to hand signal court verdict to be checked, obtained and hand to be checked
Gesture control order corresponding to gesture signal.
2. gesture identification method according to claim 1, it is characterised in that in construction basic gesture collection (10) step, institute
Multiple basic gestures are stated to meet：Each basic gesture is divided into etc. to multiple run duration sections of quantity, same sequence number when
Between in section, change between each basic gesture relative to the direction of motion of receiver, motion amplitude or range of movement, so that respectively
Minimum norm distance is maximum between the characteristic vector of individual basic gesture or makes the characteristic vector of each basic gesture orthogonal.
3. gesture identification method according to claim 1, it is characterised in that：Basic gesture off-line training (20) step
Including：
21) basic gesture is segmented：To each basic gesture, its complete procedure is divided into etc. to the multistage of duration；
22) characteristic scalar obtains：By signal characteristic discretization corresponding to every section of gesture, the signal characteristic mark of each section of gesture is obtained
Amount；
23) characteristic vector obtains：The signal characteristic scalar of whole basic gesture is combined successively, forms the signal of the basic gesture
Characteristic vector；
24) basic gesture feature database is established：The character vector of signals of each basic gesture forms basic gesture feature database.
4. gesture identification method according to claim 3, it is characterised in that：In described basic gesture segmentation (21) step,
Basic gesture division number is 5~30 sections.
5. gesture identification method according to claim 3, it is characterised in that it is specific that the characteristic scalar obtains (22) step
For：
The multiple measured value of each segment signal is averaged and used as the characteristic scalar of each segment signal, each segment signal characteristic scalar
The absolute figure of measured value only takes its symbol.
6. gesture identification method according to claim 3, it is characterised in that the characteristic scalar is obtained in (22) step,
The signal characteristic includes the variable quantity of signal amplitude, phase, signal direction of arrival or Doppler frequency shift with gesture motion.
7. gesture identification method according to claim 1, it is characterised in that：Described gesture coding (30) step includes：
31) coding mode judges：Judge gesture coding mode, encoded if non-isometric degree, then go to fault-tolerant encoding mode decision
(33) step；
32) gesture generation is combined：Combined using the basic gesture of identical quantity, form combination gesture, realize gesture command to group
Close the mapping of gesture；
33) fault-tolerant encoding mode decision：Judge fault-tolerant encoding pattern, if Error Correction of Coding pattern, then go to Error Correction of Coding generation
(35) step；
34) error detection coding generates：Error detection coding is carried out to gesture command or combination gesture, according to equal length code, then combined
Any two code character at least two code elements are different in the coding schedule of gesture error detection coding, encoded according to non-isometric degree, then hand
Either length is different or at least two code elements are different for any two code character in the coding schedule of gesture order error detection coding；
35) Error Correction of Coding generates：Error Correction of Coding is carried out to gesture command or combination gesture, according to equal length code, then combined
Any two code character at least three symbol differences in the coding schedule of gesture Error Correction of Coding, encoded according to non-isometric degree, then hand
Either length is different or at least three symbol differences for any two code character in the coding schedule of gesture order Error Correction of Coding.
8. gesture identification method according to claim 1, it is characterised in that：The hand signal to be checked obtains (40) step
In, transmission signal includes：Tone signal, two-tone signal, orthogonal frequency-division multiplex singal, frequency modulation continuous wave signal.
9. gesture identification method according to claim 1, it is characterised in that：It is described in hand signal identification (50) step
Sub- gesture identification includes：
51) sub- hand signal segmentation：To each sub- hand signal, its complete procedure is divided into the multistage for waiting duration, its segmentation
Number is segmented in (21) step that division number is identical with the basic gesture, the signal characteristic scalar of each segmentation is calculated, by elder generation
Sequential combination retains each in sub- hand signal characteristic vector into sub- hand signal characteristic vector, and according to pattern is preset afterwards
The absolute figure of individual scalar only retains its symbol；
52) signal characteristic type judges：Judge the signal characteristic type of sub- hand signal, be such as the symbol of measured value, then turn
To norm calculation (54) step；
53) angle calcu-lation：The angle of sub- hand signal characteristic vector and all characteristic vectors in basic gesture feature database is calculated, is turned
(55) step is matched to gesture；
54) norm calculation：Calculate the model of the difference of all characteristic vectors in sub- hand signal characteristic vector and basic gesture feature database
Number；
55) sub- gesture matching：Find with the angle of sub- hand signal characteristic vector in basic gesture feature database or believe with sub- gesture
The basic gesture of the Norm minimum of the difference of number characteristic vector, as the basic gesture to match with sub- hand signal.
10. gesture identification method according to claim 1, it is characterised in that：The hand signal decodes (60) step bag
Include：
61) fault-tolerant encoding mode decision：Judge gesture fault-tolerant encoding pattern, if error correction mode, then go to gesture error-correcting decoding
(63) step；
62) gesture error detection decodes：Judgement gesture identification whether there is mistake, if find mistake if stop decoding, gesture control into
For invalid operation, otherwise decode and go to gesture control order output (64) step；
63) gesture error-correcting decoding：Using optimal maximum-likelihood decoding or minimal distance docoding, correct what is occurred in gesture identification
Mistake, obtain decoding result；
64) gesture control order exports：Gesture control order corresponding with decoding result is exported.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410423086.9A CN105373215B (en) | 2014-08-25 | 2014-08-25 | Dynamic radio gesture identification method with decoding is encoded based on gesture |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410423086.9A CN105373215B (en) | 2014-08-25 | 2014-08-25 | Dynamic radio gesture identification method with decoding is encoded based on gesture |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105373215A CN105373215A (en) | 2016-03-02 |
| CN105373215B true CN105373215B (en) | 2018-01-30 |
Family
ID=55375485
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410423086.9A Active CN105373215B (en) | 2014-08-25 | 2014-08-25 | Dynamic radio gesture identification method with decoding is encoded based on gesture |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105373215B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107894830B (en) * | 2017-10-12 | 2019-07-26 | 深圳大学 | A kind of interaction input method based on acoustic perceptual, system and medium |
| WO2019144295A1 (en) * | 2018-01-23 | 2019-08-01 | 深圳市大疆创新科技有限公司 | Flight control method and device, and aircraft, system and storage medium |
| CN109189218B (en) * | 2018-08-20 | 2019-05-10 | 广州市三川田文化科技股份有限公司 | A kind of method, apparatus of gesture identification, equipment and computer readable storage medium |
| CN111474587B (en) * | 2019-01-23 | 2023-05-30 | 富士通株式会社 | Article detection method and device |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102027440A (en) * | 2008-03-18 | 2011-04-20 | 艾利普提克实验室股份有限公司 | Object and movement detection |
| CN102508547A (en) * | 2011-11-04 | 2012-06-20 | 哈尔滨工业大学深圳研究生院 | Computer-vision-based gesture input method construction method and system |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| GB0806196D0 (en) * | 2008-04-04 | 2008-05-14 | Elliptic Laboratories As | Multi-range object location estimation |
| US8907929B2 (en) * | 2010-06-29 | 2014-12-09 | Qualcomm Incorporated | Touchless sensing and gesture recognition using continuous wave ultrasound signals |
-
2014
- 2014-08-25 CN CN201410423086.9A patent/CN105373215B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102027440A (en) * | 2008-03-18 | 2011-04-20 | 艾利普提克实验室股份有限公司 | Object and movement detection |
| CN102508547A (en) * | 2011-11-04 | 2012-06-20 | 哈尔滨工业大学深圳研究生院 | Computer-vision-based gesture input method construction method and system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105373215A (en) | 2016-03-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105373215B (en) | Dynamic radio gesture identification method with decoding is encoded based on gesture | |
| KR102417138B1 (en) | Method for verifying command based on virtual code, controlled device, and apparatus for making control signal | |
| EP2996360B1 (en) | System for matching signatures on the basis of motion signature information | |
| Nakazawa et al. | Indoor positioning using a high-speed, fish-eye lens-equipped camera in visible light communication | |
| CN102209956B (en) | Cognitive error control coding for channels with memory | |
| CN105809102A (en) | Handwriting Track Recognition Method and System | |
| CN104503275A (en) | Non-contact control method and equipment based on gestures | |
| CN102263612B (en) | Error detection method used for data block and equipment thereof | |
| CN108306361B (en) | Wireless charging device and method | |
| CN107005250A (en) | Multi-thread code element changes clock code element error correction | |
| JP2016012916A5 (en) | ||
| CN107959925A (en) | A kind of smart home device assisted location method | |
| CN107402374A (en) | A kind of localization method, server and alignment system | |
| US20180232574A1 (en) | Systems and Methods for Transmitting and Receiving Data Using Machine Learning Classification | |
| Bastwesy et al. | Deep learning sign language recognition system based on wi-fi csi | |
| CN114328908A (en) | Question and answer sentence quality inspection method and device and related products | |
| CN105468211A (en) | Data transmission method, system and device between capacitive stylus and touch screen | |
| CN106028001B (en) | A kind of optical positioning method and device | |
| CN102200829A (en) | Body movement recognition device used for virtual reality input | |
| CN105222771B (en) | A kind of indoor article localization method and user terminal | |
| CN104240492A (en) | Method, device and terminal for debugging infrared remote control signals | |
| CN105183726A (en) | Method and system for determining user similarity | |
| CN101441811A (en) | Infrared remote controller and method for transmitting signal thereof and infrared remote-control system | |
| CN109541528A (en) | A kind of electromagnetic tracking system identified based on target state and propagation medium | |
| Tu et al. | Mechanical arm teleoperation control system by dynamic hand gesture recognition based on kinect device |
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 |