CN109460716A - A kind of sign language wireless-identification device and method - Google Patents
A kind of sign language wireless-identification device and method Download PDFInfo
- Publication number
- CN109460716A CN109460716A CN201811218349.7A CN201811218349A CN109460716A CN 109460716 A CN109460716 A CN 109460716A CN 201811218349 A CN201811218349 A CN 201811218349A CN 109460716 A CN109460716 A CN 109460716A
- Authority
- CN
- China
- Prior art keywords
- sign language
- wireless
- microprocessor
- wireless signal
- recognition
- 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.)
- Withdrawn
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/20—Movements or behaviour, e.g. gesture recognition
- G06V40/28—Recognition of hand or arm movements, e.g. recognition of deaf sign language
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/10—Image acquisition
- G06V10/12—Details of acquisition arrangements; Constructional details thereof
- G06V10/14—Optical characteristics of the device performing the acquisition or on the illumination arrangements
- G06V10/143—Sensing or illuminating at different wavelengths
Abstract
The present invention provides a kind of sign language wireless-identification device and method, belongs to medical treatment & health and information technology field.The device does not need device that user carries any equipment, only passing through transmitting radio wave and measure and analyze character of reflection wave realization Sign Language Recognition to be a kind of.The device continuously emits wireless signal, receives the wireless signal influenced by sign language movement, sign language feature is extracted from wireless signal amplitude and phase information based on deep learning network, and carry out classification to sign language feature based on classifier to realize Sign Language Recognition.The device of the invention carries any equipment without user, is not related to privacy leakage problem, can use under the complex scenes such as dark, dense smoke, through walls, and cost is relatively low.It is exchanged in assisting hearing impedient people, and there is good application prospect in the fields such as special operation workers gesture remote control operation.
Description
Technical field
The invention belongs to medical treatment & health and information technology field, it is related to a kind of sign language wireless-identification device and method.Specifically
For it is a kind of do not need user carry any equipment, only by transmitting radio wave and measure with analysis the sign language of human body difference movement
The device of character of reflection wave realization Sign Language Recognition.
Background technique
It is counted according to the nearest national census, there are 20,750,000 dysaudia personages in China at present.Sign language is as hearing
The language of the daily exchange of impedient people will be exchanged and be incorporated for dysaudia personage if sign language can be identified automatically
Society provides basic guarantee, effectively promotes reaching for the fine vision of harmonious society.
Currently, researcher, which has attempted the method based on image procossing and based on wearable device, realizes Sign Language Recognition.
Related work based on image procossing, such as " J.Huang, W.Zhou, H.Li, and W.Li.Sign Language
Recognition using 3D Convolutional Neural Networks.In IEEE ICME 2015,pp.1–
6. ", " Hu Zhangfang, Luo Yuan, Zhang Yi, Yang Lin, Xi Bing, static manual alphabet identifying system and method based on Kinect sensor,
Chinese invention patent, patent No. CN201410191394.3 ", such method capture human body gesture motion by Kinect camera
Video realizes sign language action recognition based on image processing method later.Although such method can obtain satisfactory identification
Precision, however, it is desirable to using camera be aligned dysaudia personage, this cause this method be not easy in daily life at any time with
Ground uses, meanwhile, it is related to privacy leakage problem.Related work based on wearable device, such as " B.Fang, J.Co, and
M.Zhang.DeepASL:Enabling Ubiquitous and Non-Intrusive Word and Sentence-Level
Sign Language Translation.In ACM SenSys 2017, pp.1-13. ", " WangBen, Jiang little Hua, Luo Hong appoint
Red horse, Zhang Jianjie, Xu Kui, Chen Yanlin, straight space, Guo Wenwei, for acquiring the gloves of Sign Language Recognition data, Chinese invention is special
Benefit, patent No. CN201410410413.7 ", such method utilize and are equipped with the various sensors such as acceleration, inclination angle, gyroscope
Intelligent glove captures the various motion informations of sign language movement, realizes sign language action recognition based on motion information analysis method later.
Although such method can obtain higher accuracy of identification, however, it is desirable to dedicated Intelligent glove is configured for dysaudia personage,
Higher cost and inconvenient for use.
Summary of the invention
The purpose of the present invention is overcoming the deficiencies of existing technologies, provide it is a kind of do not need user carry any equipment, only
Pass through transmitting radio wave and measures the device for realizing Sign Language Recognition with analysis character of reflection wave.Compared with the prior art, this hair
Bright device carries any equipment without user, is not related to privacy leakage problem, can be in the complex scenes such as dark, dense smoke, through walls
Lower use, and cost is relatively low.
Technical solution of the present invention:
A kind of sign language wireless-identification device, the sign language wireless-identification device include wireless transmitter 1,2 and of wireless receiver
Microprocessor 3;The wireless transmitter 1 continuously emits wireless signal;The wireless receiver 2 and 3 phase of microprocessor
Even, wireless receiver 2 receives the wireless signal that wireless transmitter 1 emits, by wireless signal amplitude and phase information after demodulation
It is transmitted to microprocessor 3;Sign Language Recognition signal processing software, Sign Language Recognition signal processing software are configured in the microprocessor 3
Low pass filtered is carried out to the wireless signal amplitude and phase information come from the transmission of wireless receiver 2 and involves wavelet transform process, and base
Sign language feature is extracted from wireless signal amplitude and phase information in deep learning network, and sign language feature is carried out based on classifier
Classification is to realize Sign Language Recognition.
The carrier frequency of the wireless transmitter 1 is 1GHz or more, and transmission power can be adjusted in 0dBm between 10dBm,
Configure the omnidirectional antenna that gain is greater than 3dB;
Carrier frequency, transmission power and the modulation system of the wireless receiver 2 are identical as wireless transmitter 1, receive
Sensitivity is -100dBm hereinafter, configuration gain is greater than the omnidirectional antenna of 3dB, by network interface or USB interface and microprocessor 3 it
Between realize information transmission;
The CPU frequency of the microprocessor 3 is greater than 2GHz, and memory size is greater than 8G byte.
A kind of sign language wireless identification method, this method include off-line training and two stages of online recognition, off-line training rank
Section acts to learn the parameter of deep learning network and classifier according to known sign language, and the online recognition stage is seen according to current
The sign language feature measured realizes Sign Language Recognition, the specific steps are as follows:
1) off-line training step
(1.1) human body executes a certain known sign language movement, meanwhile, wireless transmitter 1 continuously emits radio wave, wirelessly
Receiver 2 receives the wireless signal influenced by human body sign language movement, passes wireless signal amplitude and phase information after demodulation
It is sent to microprocessor 3;
(1.2) the Sign Language Recognition signal processing software in microprocessor 3 uses low-pass filter to wireless signal width first
Degree is filtered respectively with phase information, is then believed using wavelet transform filtered amplitude information and phase
Breath is handled, and the temporal frequency two-dimensional matrix information of amplitude and phase information is obtained;Deep learning network is later with time frequency
Rate two-dimensional matrix information is input, carries out convolution, down-sampling and nonlinear transformation to information and operates, it is special to obtain one-dimensional sign language
Sign;Last classifier is input with one-dimensional sign language feature, output element number be equal to the one-dimensional sign language type of sign language species number to
Amount.
(1.3) it is acted according to known true sign language and the sign language of identification acts, using error backpropagation algorithm pair
The parameter of deep learning network and classifier is updated study;
(1.4) step (1.1) to (1.3) are repeated, until the parameter of deep learning network and classifier remains unchanged, from
The line training stage finishes.
2) the online recognition stage
(2.1) human body executes a certain unknown sign language movement, meanwhile, wireless transmitter 1 continuously emits radio wave, wirelessly
Receiver 2 receives the wireless signal influenced by human body sign language movement, passes wireless signal amplitude and phase information after demodulation
It is sent to microprocessor 3;
(2.2) progress of Sign Language Recognition signal processing software and step (1.2) identical processing that microprocessor 3 configures, most
Real-time Sign Language Recognition is realized eventually.
Beneficial effects of the present invention: it can provide a kind of only real by transmitting radio wave and measurement and analysis character of reflection wave
The device of existing Sign Language Recognition, the device carry any equipment without user, are not related to privacy leakage problem, can dark, dense smoke,
It is used under equal complex scenes through walls, and cost is relatively low.It is exchanged in assisting hearing impedient people, special operation workers gesture remote control
There is good application prospect in the fields such as operation.
Detailed description of the invention
Fig. 1 is the system structure diagram of apparatus of the present invention.
In figure: 1 wireless transmitter;2 wireless receivers;3 microprocessors.
Fig. 2 is the working principle diagram of the method for the present invention.
Specific embodiment
Specific implementation of the invention is specifically elaborated below with reference to technical solution and attached drawing.
Embodiment uses system structure diagram shown in FIG. 1, and System Working Principle is as shown in Figure 2.The composition of device is as follows:
WiFi chip Intel5300 of the wireless transmitter 1 using work in 5GHz realizes that transmission power 10dBm assembles 3 3dB gains
Omnidirectional antenna;Wireless receiver 2 uses work in WiFi chip Intel5300 and the USB3.0 chip of 5GHz
CY7C68013A, receiver sensitivity -105dBm assemble the omnidirectional antenna of 3 3dB gains, by USB3.0 interface to micro- place
It manages device 3 and transmits data;Microprocessor 3 runs the sign language write based on Matlab in PC machine using X240s portable PC machine is associated
Identification signal processing software;Sign Language Recognition signal processing software is using the chebyshev low-pass filter of cutoff frequency 10Hz to nothing
Line signal amplitude is filtered respectively with phase information, and 6 grades of wavelet transforms is used to believe length for 1024 amplitude
Breath and phase information carry out processing and generate temporal frequency two-dimensional matrix information;Deep learning network uses the depth of 3-tier architecture
Convolutional network, is input with temporal frequency two-dimensional matrix information, and every layer of structure carries out convolution, down-sampling, Yi Jifei to information
Linear transformation operation, obtains one-dimensional sign language feature;Classifier uses Softmax classifier, is input with one-dimensional sign language feature, defeated
Element number is equal to the one-dimensional sign language type vector of sign language species number out.
Test shows that sign language wireless-identification device correct recognition rata when identifying 50 kinds of sign language movements is 92%, is knowing
Correct recognition rata is 98% when other 20 kinds of sign languages act.
Claims (6)
1. a kind of sign language wireless-identification device, which is characterized in that the sign language wireless-identification device includes wireless transmitter (1), nothing
Line receiver (2) and microprocessor (3);The wireless transmitter (1) continuously emits wireless signal;The wireless receiving
Machine (2) is connected with microprocessor (3), and wireless receiver (2) receives the wireless signal of wireless transmitter (1) transmitting, after demodulation
Wireless signal amplitude and phase information are transmitted to microprocessor (3);In the microprocessor (3) at configuration Sign Language Recognition signal
Manage software, Sign Language Recognition signal processing software to from wireless receiver (2) transmission come wireless signal amplitude and phase information into
Row low pass filtered involves wavelet transform process, and extracts sign language from wireless signal amplitude and phase information based on deep learning network
Feature carries out classification to sign language feature based on classifier to realize Sign Language Recognition.
2. a kind of sign language wireless-identification device according to claim 1, which is characterized in that the wireless transmitter (1)
Carrier frequency is 1GHz or more, and transmission power can be adjusted in 0dBm between 10dBm, and configuration gain is greater than the omnidirectional antenna of 3dB.
3. a kind of sign language wireless-identification device according to claim 1 or 2, which is characterized in that the wireless receiver (2)
Carrier frequency, transmission power and modulation system it is identical as wireless transmitter (1), receiving sensitivity be -100dBm hereinafter,
The omnidirectional antenna that gain is greater than 3dB is configured, by the transmission for realizing information between network interface or USB interface and microprocessor (3).
4. a kind of sign language wireless-identification device according to claim 1 or 2, which is characterized in that the microprocessor (3)
CPU frequency is greater than 2GHz, and memory size is greater than 8G byte.
5. a kind of sign language wireless-identification device according to claim 3, which is characterized in that the CPU of the microprocessor (3)
Dominant frequency is greater than 2GHz, and memory size is greater than 8G byte.
6. using a kind of any sign language wireless identification method of claim 1-5, which is characterized in that this method includes offline
Trained and two stages of online recognition, off-line training step act to learn deep learning network and divide according to known sign language
The parameter of class device, online recognition stage realize Sign Language Recognition according to the sign language feature that Current observation arrives, the specific steps are as follows:
1) off-line training step
(1.1) human body executes a certain known sign language movement, meanwhile, wireless transmitter (1) continuously emits radio wave, wirelessly connects
Receipts machine (2) receives the wireless signal influenced by human body sign language movement, passes wireless signal amplitude and phase information after demodulation
It is sent to microprocessor (3);
(1.2) the Sign Language Recognition signal processing software in microprocessor (3) uses low-pass filter to wireless signal amplitude first
It is filtered respectively with phase information, then using wavelet transform to filtered amplitude information and phase information
It is handled, obtains the temporal frequency two-dimensional matrix information of amplitude and phase information;Deep learning network is later with temporal frequency
Two-dimensional matrix information is input, carries out convolution, down-sampling and nonlinear transformation to information and operates, obtains one-dimensional sign language feature;
Last classifier is input with one-dimensional sign language feature, and output element number is equal to the one-dimensional sign language type vector of sign language species number;
(1.3) it is acted according to known true sign language and the sign language of identification acts, using error backpropagation algorithm to depth
Learning network and the parameter of classifier are updated study;
(1.4) step (1.1) to (1.3) are repeated, it is offline to instruct until the parameter of deep learning network and classifier remains unchanged
The white silk stage finishes;
2) the online recognition stage
(2.1) human body executes a certain unknown sign language movement, meanwhile, wireless transmitter (1) continuously emits radio wave, wirelessly connects
Receipts machine (2) receives the wireless signal influenced by human body sign language movement, passes wireless signal amplitude and phase information after demodulation
It is sent to microprocessor (3);
(2.2) progress of Sign Language Recognition signal processing software and step (1.2) identical processing of microprocessor (3) configuration, finally
Realize real-time Sign Language Recognition.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811218349.7A CN109460716A (en) | 2018-10-19 | 2018-10-19 | A kind of sign language wireless-identification device and method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811218349.7A CN109460716A (en) | 2018-10-19 | 2018-10-19 | A kind of sign language wireless-identification device and method |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109460716A true CN109460716A (en) | 2019-03-12 |
Family
ID=65607856
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811218349.7A Withdrawn CN109460716A (en) | 2018-10-19 | 2018-10-19 | A kind of sign language wireless-identification device and method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109460716A (en) |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20150009116A1 (en) * | 2013-07-02 | 2015-01-08 | Disney Enterprises, Inc. | Facilitating gesture-based association of multiple devices |
CN104504396A (en) * | 2014-12-18 | 2015-04-08 | 大连理工大学 | Method for recognizing position state of human body by utilizing natural environment wireless signal |
CN104503575A (en) * | 2014-12-18 | 2015-04-08 | 大连理工大学 | Method for designing low-power-consumption gesture recognition circuit device |
CN105573498A (en) * | 2015-12-15 | 2016-05-11 | 西北大学 | Gesture recognition method based on Wi-Fi signal |
CN106407905A (en) * | 2016-08-31 | 2017-02-15 | 电子科技大学 | Machine learning-based wireless sensing motion identification method |
CN107067031A (en) * | 2017-03-29 | 2017-08-18 | 西北大学 | A kind of calligraphy posture automatic identifying method based on Wi Fi signals |
CN107463887A (en) * | 2017-07-20 | 2017-12-12 | 四川长虹电器股份有限公司 | Train driver gesture intelligence inspection system and intelligent inspection method |
CN108416974A (en) * | 2018-01-23 | 2018-08-17 | 中国矿业大学 | Autoalarm based on radio channel status information and method |
-
2018
- 2018-10-19 CN CN201811218349.7A patent/CN109460716A/en not_active Withdrawn
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20150009116A1 (en) * | 2013-07-02 | 2015-01-08 | Disney Enterprises, Inc. | Facilitating gesture-based association of multiple devices |
CN104504396A (en) * | 2014-12-18 | 2015-04-08 | 大连理工大学 | Method for recognizing position state of human body by utilizing natural environment wireless signal |
CN104503575A (en) * | 2014-12-18 | 2015-04-08 | 大连理工大学 | Method for designing low-power-consumption gesture recognition circuit device |
CN105573498A (en) * | 2015-12-15 | 2016-05-11 | 西北大学 | Gesture recognition method based on Wi-Fi signal |
CN106407905A (en) * | 2016-08-31 | 2017-02-15 | 电子科技大学 | Machine learning-based wireless sensing motion identification method |
CN107067031A (en) * | 2017-03-29 | 2017-08-18 | 西北大学 | A kind of calligraphy posture automatic identifying method based on Wi Fi signals |
CN107463887A (en) * | 2017-07-20 | 2017-12-12 | 四川长虹电器股份有限公司 | Train driver gesture intelligence inspection system and intelligent inspection method |
CN108416974A (en) * | 2018-01-23 | 2018-08-17 | 中国矿业大学 | Autoalarm based on radio channel status information and method |
Non-Patent Citations (3)
Title |
---|
何文锋: "基于WIFI的手势识别研究", 《中国优秀博硕士学位论文全文数据库(硕士)_信息科技辑》 * |
刘熠辰等: "基于雷达技术的手势识别", 《中国电子科学研究院学报》 * |
周戈: "基于Wifi信号的手势识别技术研究", 《中国优秀博硕士学位论文全文数据库(硕士)_信息科技辑》 * |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Gu et al. | Paws: Passive human activity recognition based on wifi ambient signals | |
Khan et al. | A deep learning framework using passive WiFi sensing for respiration monitoring | |
US20200380178A1 (en) | Tracking safety conditions of an area | |
CN105590031A (en) | Data processing device applied to intelligent medical facility | |
CN108287002A (en) | A kind of remote environment-monitoring device | |
CN106557744B (en) | Wearable face identification device and implementation method | |
CN108073910A (en) | For generating the method and apparatus of face characteristic | |
Zhang et al. | Gate-ID: WiFi-based human identification irrespective of walking directions in smart home | |
Xiao et al. | Motion-fi: Recognizing and counting repetitive motions with passive wireless backscattering | |
Xu et al. | Attention-based gait recognition and walking direction estimation in wi-fi networks | |
CN108418867A (en) | A kind of method of environmental monitoring preventing human interference | |
Samadh et al. | Indoor localization based on channel state information | |
CN108534822A (en) | A kind of remote environment-monitoring device | |
Jia et al. | BeAware: Convolutional neural network (CNN) based user behavior understanding through WiFi channel state information | |
CN108344446A (en) | A kind of method of environmental monitoring preventing human interference | |
CN108414008A (en) | A kind of city environmental quality monitoring system | |
Zhang et al. | Human identification using WiFi signal | |
Yang et al. | TARF: Technology-agnostic RF sensing for human activity recognition | |
CN104013401B (en) | A kind of human body electroencephalogram's signal and action behavior signal synchronous collection system and method | |
CN109460716A (en) | A kind of sign language wireless-identification device and method | |
Kabir et al. | CSI-DeepNet: A Lightweight Deep Convolutional Neural Network Based Hand Gesture Recognition System Using Wi-Fi CSI Signal | |
US20180225988A1 (en) | Sign language gesture determination systems and methods | |
CN208492052U (en) | A kind of wrist strap | |
KR20110032429A (en) | Apparatus and method for recognzing gesture in mobile device | |
Ding et al. | Device-free location-independent human activity recognition via few-shot learning |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
WW01 | Invention patent application withdrawn after publication |
Application publication date: 20190312 |
|
WW01 | Invention patent application withdrawn after publication |