阅读说明：本技术 一种基于WiFi手势识别的智能家居管理系统 (Smart home management system based on WiFi gesture recognition ) 是由 谭吉峰 张杨 夏思远 余粞淼 别文宇 于 2020-05-19 设计创作，主要内容包括：本发明一种基于WiFi手势识别的智能家居管理系统,包括WiFi发射模块、信道模块、WiFi接收模块、手势识别模块以及家居控制模块；所述WiFi发射模块包括WiFi发射器和WiFi信号；所述信道模块包括室内环境调制及手势调制；所述WiFi接收模块包括WiFi信号接收机、滤波电路及放大电路、信号采集模块；所述手势识别模块包括信号剪切、信号拼接、时频分析算法、线谱提取技术、动态时间规整,特征识别,所述家居控制模块(5)包括指令传输和相应方式控制；本发明通过截取WiFi信号的长前导,利用时频谱分析、线谱提取、支持向量机分类以及动态时间规整算法完成手势的识别并进行家居管理控制。从而快速、无声的对家具进行管控。避免语音控制以及手机APP控制造成的影响。(The invention relates to an intelligent home management system based on WiFi gesture recognition, which comprises a WiFi transmitting module, a channel module, a WiFi receiving module, a gesture recognition module and a home control module, wherein the WiFi transmitting module is used for transmitting a signal to the signal channel module; the WiFi transmitting module comprises a WiFi transmitter and a WiFi signal; the channel module comprises indoor environment modulation and gesture modulation; the WiFi receiving module comprises a WiFi signal receiver, a filter circuit, an amplifying circuit and a signal acquisition module; the gesture recognition module comprises signal shearing, signal splicing, a time-frequency analysis algorithm, a line spectrum extraction technology, dynamic time warping and feature recognition, and the home control module (5) comprises instruction transmission and corresponding mode control; the invention intercepts the long preamble of the WiFi signal, completes gesture recognition and household management control by utilizing time spectrum analysis, line spectrum extraction, support vector machine classification and dynamic time warping algorithm. Therefore, the furniture is rapidly and silently controlled. Avoid the influence that speech control and cell-phone APP control caused.)

1. An intelligent home management system based on WiFi gesture recognition is characterized by comprising a WiFi transmitting module (1), a channel module (2), a WiFi receiving module (3), a gesture recognition module (4) and a home control module (5); the WiFi transmission module (1) comprises a WiFi transmitter (10) and a WiFi signal (11); the channel module (2) comprises an indoor environment modulation (20) and a gesture modulation (21); the WiFi receiving module (3) comprises a WiFi signal receiver (31), a filter circuit (32), an amplifying circuit (33) and a signal acquisition module (34); the gesture recognition module (4) comprises signal shearing (40), signal splicing (41), a time-frequency analysis algorithm (42), a line spectrum extraction technology (43), dynamic time warping (44) and feature recognition (45), and the home control module (5) comprises instruction transmission (50) and corresponding mode control (51).

2. The smart home management system based on WiFi gesture recognition of claim 1, wherein the command transmission (50) comprises: corresponding gesture judgment for judging the start and the end of home control and home control command transmission opening and closing; the system comprises gesture group identification used for judging the home to be controlled and home management instruction transmission corresponding to the gesture group.

3. The smart home management system based on WiFi gesture recognition according to claim 1 or 2, wherein the corresponding mode control (51) comprises a direct control switch mode (510), a direct control duty cycle mode (511) and a WiFi remote control mode (512).

4. A smart home management system based on WiFi gesture recognition according to claim 1 or 2, characterized by that the WiFi signal (11) is based on 802.11a protocol.

5. A smart home management system based on WiFi gesture recognition according to claim 3 characterized by that the WiFi signal (11) is based on 802.11a protocol.

6. The smart home management system based on WiFi gesture recognition according to claim 1 or 2, wherein the WiFi transmitter (10) uses an indoor wireless router; the WiFi signal receiver (31) adopts a radio frequency signal receiving module to collect WiFi signals, and the circuits condition the signals through a high-frequency high-power-amplifier signal conditioning hardware filter circuit (32) and an amplifying circuit (33), so that high signal-to-noise ratio signals are obtained; the method comprises the steps that WiFi signals which contain gesture features and are subjected to hardware conditioning are collected and transmitted to a gesture recognition module (4), and the gesture recognition module (4) comprises signal shearing (40), signal splicing (41), a time-frequency analysis algorithm (42), line spectrum extraction (43) and feature recognition (44); the household control module (5) is used for carrying out regional control on household.

7. The smart home management system based on WiFi gesture recognition as claimed in claim 1 or 2, wherein the gesture modulation comprises a start gesture, a stop gesture and a data gesture.

Technical Field

The invention relates to an intelligent home management system, in particular to an intelligent home management system based on WiFi gesture recognition.

Background

The WiFi signal is as a radio wave operating in the 2.4GHz and 5.8GHz frequency bands. Due to the characteristics of high frequency, small wavelength, sufficient bandwidth and the like, the method is extremely suitable for the process of transmitting a large amount of data, and is widely applied to the field of short-distance wireless communication. Meanwhile, with the development of pattern recognition and man-machine interaction technologies, the strong capability of the WiFi signals in recognition is gradually mined out.

Gesture recognition is a subject of computer science and language technology, with the aim of recognizing human gestures by mathematical algorithms. Gestures may originate from any body motion or state, but typically originate from the face or hands. Can be viewed as a way of computationally solving human language, thus building a richer bridge between machines and humans than the original text user interface. The initial gesture recognition mainly uses a machine device to directly detect the angle and the spatial position of each joint of the arm. Most of these devices connect the computer system and the user to each other through a wired technology, so that the gesture information of the user can be transmitted to the recognition system completely and without errors, and typical devices of the devices are data gloves and the like. Thereafter, the optical marking method replaces a data glove to wear optical marks on the human hand, and changes of the position of the human hand and fingers can be transmitted to a system screen through infrared rays. The mainstream method of the existing gesture recognition technology is gesture image recognition. The realization of the gesture recognition technology firstly needs gesture segmentation as a key step in the gesture recognition process, and the effect of the gesture segmentation directly influences the next gesture analysis and the final gesture recognition. The most common gesture segmentation methods at present mainly include gesture segmentation based on monocular vision and gesture segmentation based on stereoscopic vision. After the gesture segmentation is completed, the key technology of the gesture recognition system is completed, and the gesture analysis is completed. Through the gesture analysis, the shape characteristics or the motion trail of the gesture can be obtained. The shape and motion trajectory of the gesture are important features in dynamic gesture recognition, and have a direct relation with the meaning expressed by the gesture. Finally, the analyzed gestures need to be recognized, and the recognition method is a process of classifying tracks (or points) in the model parameter space into a certain subset in the space, and the process comprises static gesture recognition and dynamic gesture recognition.

When WiFi passes through a changing gesture, the transmission characteristics such as amplitude, phase and power of the signal are affected to some extent, and the effect is generated based on the moving characteristics of the dynamic gesture. This means that the WiFi signal is modulated by the gesture in a manner that the modulated signal contains the movement characteristics information of the gesture it is traversing. The gesture recognition can be realized by demodulating the information by a certain means.

The first mor code facility equipment maintenance (Beijing) limited, 4 months 2016, proposed a WiFi smart gesture sensor (CN 205193495U); the Samsung Enterprise development (Shanghai) limited, 7 months in 2016, proposes a gesture unlocking device based on WiFi and a control method (CN105809787A) thereof, thereby realizing the practical application of the technology of gesture recognition by utilizing WiFi signals; a light-adjustable LED intelligent wireless controller (CN205546080U) is provided by the national optical and electrical technology corporation of the Ji-Gekko, 8 months in 2016, and the application of a WiFi gesture recognition technology is improved to be controlled by multiple options with the same parameter through simple switch control; the SUZHOU research institute of Chinese science and technology university in 1 month 2019 proposes a gesture control human-computer interaction system (CN105807935B) based on WiFi.

The establishment of the gesture recognition system has the advantages that due to the existence of certain background noise in the signal acquisition process, such as interference of non-target objects, the signals are required to be subjected to smoothing, sharpening and binarization processing through data preprocessing so as to extract more obvious gesture modulation signals; in order to offset the huge reflection component from the wall, the Wi-Vi system of the massachusetts institute of technology first estimates the whole system channel under the condition of no moving target, and then uses the established channel model to eliminate the interference of the received short signal under the condition of moving target, thereby eliminating the wall reflection component.

The intelligent home is also called as an intelligent house, and is characterized in that advanced computer technology, network communication technology, intelligent cloud control, comprehensive wiring technology and medical electronic technology are utilized to integrate individual requirements according to the principle of human engineering, various subsystems related to home life, such as security protection, light control, curtain control, gas valve control, information household appliances, scene linkage, floor heating, health care, epidemic prevention, security protection and the like, are organically combined together, and people-oriented brand-new home life experience is realized through networked comprehensive intelligent control and management. The control method of the current smart home control system mainly includes voice control and mobile phone control. Indoor and remote convenient control can be realized.

For an intelligent home control system, how to quickly and accurately obtain a control instruction and effectively respond to the control instruction is a performance which needs to be improved all the time. How to simply expand the household types is also a subject to be researched. According to the invention, the indoor WiFi is used as the carrier signal to transmit the command to the gesture signal, so that the furniture control command can be conveniently obtained. Meanwhile, the WiFi-based gesture recognition is lower in cost, higher in accuracy and better in confidentiality compared with other recognition technologies.

Disclosure of Invention

The invention mainly aims to identify the gestures of indoor moving personnel by utilizing indoor WiFi signals and properly adjust corresponding home by utilizing related gestures, thereby providing the intelligent home management system based on WiFi gesture identification, which has high speed, low energy consumption and accurate control.

The invention is realized by the following steps:

an intelligent home management system based on WiFi gesture recognition is characterized by comprising a WiFi transmitting module, a channel module, a WiFi receiving module, a gesture recognition module and a home control module; the WiFi transmitting module comprises a WiFi transmitter and a WiFi signal; the channel module comprises indoor environment modulation and gesture modulation; the WiFi receiving module comprises a WiFi signal receiver, a filter circuit, an amplifying circuit and a signal acquisition module; the gesture recognition module comprises signal shearing, signal splicing, a time-frequency analysis algorithm, a line spectrum extraction technology, dynamic time warping and feature recognition, and the home control module comprises instruction transmission and corresponding mode control.

The invention also includes such features:

the instruction transmission comprises: corresponding gesture judgment for judging the start and the end of home control and home control command transmission opening and closing; the system comprises gesture group identification used for judging the home to be controlled and home management instruction transmission corresponding to the gesture group;

the corresponding mode control comprises a direct control switch mode, a direct control duty ratio mode and a WiFi remote control mode;

the WiFi signal is based on 802.11a protocol;

wherein the WiFi transmitter uses an indoor wireless router; the WiFi signal receiver adopts a radio frequency signal receiving module to collect WiFi signals, and the circuits condition the signals through a high-frequency high-power-amplifier signal conditioning hardware filter circuit and an amplifying circuit, so that high signal-to-noise ratio signals are obtained; the method comprises the steps that WiFi signals which contain gesture features and are subjected to hardware conditioning are collected and transmitted to a gesture recognition module, and the gesture recognition module comprises signal shearing and signal splicing, a time-frequency analysis algorithm, line spectrum extraction and feature recognition; the home control module is used for carrying out regional control on home;

the gesture modulation includes a start gesture, a stop gesture, and a data gesture.

The invention has the beneficial effects that:

the indoor WiFi is adopted to perform gesture recognition and subsequent technical operation under the condition of not influencing network connection, so that the cost is lower, the installation is convenient, and a signal transmitting system does not need to be additionally developed;

in the signal preprocessing process, a background noise elimination means is adopted, so that the interference of environmental noise on the gesture signal is effectively reduced, and the gesture recognition accuracy is improved;

the room, home and home adjustable parameters are numbered and controlled through the gesture group. Through the arrangement of the multi-digit array, a gesture is replaced to indicate the adjustment of a parameter, so that resources occupied by a large amount of learning of the gesture can be effectively avoided;

compared with a voice control mode of an intelligent sound box, the home control is performed by utilizing gestures, so that the security is better, the accuracy is higher, and the situation that the home control cannot be distinguished due to the loss of syllables or characters in the voice control does not occur;

the gesture group needs the starting gesture and the ending gesture to support, the condition of mistaken touch in the normal activity process of indoor personnel can be effectively avoided, and the stability is higher.

Drawings

FIG. 1 is a schematic diagram of the system of the present invention;

FIG. 2 is a diagram of a WiFi signal structure;

FIG. 3 is a schematic diagram of a gesture recognition model comparison module;

FIG. 4 is a schematic diagram of a control system;

FIG. 5 is a simulation diagram of a dynamic time warping algorithm.

Detailed Description

The present invention will be described in detail with reference to specific embodiments.

The technical scheme adopted by the invention for solving the technical problem is as follows: a WiFi-based gesture recognition judgment instruction and a home management and control technology are adopted;

mainly comprises a gesture recognition technology and a home control technology;

the gesture recognition technology comprises a WiFi signal transmitting module 1, a signal receiving and processing module 3 and a gesture feature analysis module 4;

the WiFi signal transmitting module mainly comprises:

an indoor WiFi transmitter 10 for WiFi signal transmission;

a WiFi signal 11 containing protocols and data that can be used to connect to the network and used to be gesture modulated to contain gesture characteristics;

the signal receiving and processing module 3 of the present invention mainly comprises:

a signal receiver (radio frequency signal receiving module 30) for receiving WiFi signals;

a filter circuit 31 and an amplifier circuit 32 for conditioning the received WiFi signal containing the gesture feature. The interference of out-of-band noise can be effectively reduced, and signals can be acquired as high as possible;

an analog/digital conversion module 33 for acquiring the signal. The WiFi signals which contain gesture characteristics and are subjected to hardware conditioning are collected and transmitted to a gesture characteristic analysis module 4;

the gesture feature analysis module 4 of the present invention mainly includes:

a signal segment cutting method 40 and a signal segment splicing technique 41 for obtaining a gesture signal;

an envelope method for smoothing the acquired gesture signal. Since the signals analyzed by the gesture recognition technology are mainly long leading parts of the WiFi signals, the signal dimension is high and the spliced parts are not smooth. An envelope method is adopted to reduce the sampling rate and stabilize the signal;

a time-frequency analysis algorithm 42 for obtaining the time-frequency spectrum. Short-time Fourier transform with high speed, short-time music algorithm or wavelet transform with high accuracy and the like can be adopted;

a line spectrum extraction technique 43 for simplifying the dimension of the time-frequency analysis result from the matrix into a curve to obtain a gesture feature;

the dynamic time warping algorithm 44 is used for reducing the recognition accuracy rate caused by the change speed of the gesture, and the signals to be recognized and the models are in a similar state for comparison by distorting the time dimension;

a feature recognition algorithm 45 for determining gestures. Comparing the gesture features with the established model, and determining the gesture according to the similarity;

the home control module in the invention comprises:

instruction transmission 50 for the selected home and its corresponding adjustments. Obtaining corresponding numbers of the corresponding household appliances through a group of gestures to form an array, and selecting corresponding household appliances, parameters required to be adjusted for the household appliances and corresponding changes of the parameters;

a control mode 51 for controlling the furniture;

the signal preprocessing part comprises:

background noise cancellation to improve accuracy;

the received modulation of the WiFi signal in the channel mainly includes indoor environment modulation 20 and gesture modulation 21. Wherein the signal modulated by the fixed home is background noise;

the background noise can be eliminated, indoor environment signals with a certain length can be collected firstly after the system is installed, and the indoor environment modulation signals can be reduced in the signal processing process to improve the signal-to-noise ratio, so that more obvious gesture modulation signals can be obtained;

establishing a support vector machine classification model for classifying gestures;

the transmission 50 of the household control instruction comprises the following steps;

corresponding gesture judgment for judging the start and the end of home control and home control command transmission opening and closing;

the system comprises gesture group identification used for judging the home to be controlled and home management instruction transmission corresponding to the gesture group;

three methods 51 for controlling home in the present invention include;

a direct control switch mode 510, a direct control duty cycle mode 511, and a WiFi remote control mode 512;

the direct control switch mode 510 is primarily operative;

only one parameter can be controlled and provided with two judgment results such as back locking and unlocking of a door, switching of an illuminating lamp and an alarm system;

the false alarm cost of the alarm system is high, so that the corresponding gesture group is more complex than other household control gestures and has extremely low misjudgment probability;

the direct control duty cycle manner 511 is mainly used;

only one parameter can be controlled and continuously changed, such as brightness of a desk lamp, temperature of a refrigerator and the like;

the WiFi remote control mode 512 is mainly used;

the home with more control parameters mainly comprises complex homes such as televisions, sound equipment, air conditioners and the like;

in the present invention the WiFi signal 11 is based on the 802.11a protocol;

in the invention, the gesture signal segment is an 802.11 protocol long leading part of a WiFi signal;

the gesture signal segment cutting method 40 and the signal cutting segment splicing technology 41 in the invention are methods for accurately intercepting and continuously splicing gesture signal segments from WiFi;

the envelope method adopts a sliding window to obtain the average power of signals in a signal window and has a certain overlapped moving window so as to obtain smooth signals containing gesture characteristics, and reduces the data volume of the signals and the calculation load;

fig. 1 is a schematic structural diagram of an intelligent home management system based on WiFi gesture recognition according to an embodiment of the present invention, and the overall system is divided into several main parts, such as a WiFi transmitting module 1, a channel module 2, a WiFi receiving module 3, a gesture recognition module 4, and a home control module 5. Wherein the WiFi transmitter 10 uses an indoor wireless router; the WiFi receiver 30 adopts a radio frequency signal receiving module to collect WiFi signals, and a circuit conditions the signals through high-frequency high-power-amplifier signal conditioning hardware 31 and 32, so that high signal-to-noise ratio signals are obtained; the gesture recognition module 4 comprises signal cutting 40 and signal splicing 41 technologies, a time-frequency analysis algorithm 42, line spectrum extraction 43, feature recognition 44 and other main parts; the household control module 5 is used for performing regional control on the household.

The WiFi signal used in the present invention is based on 802.11 protocol, and its physical layer is divided into three parts as shown in fig. 2. The first part is a preamble, which is mainly used for realizing functions of carrier sense, channel estimation, symbol synchronization, frequency compensation and the like at a receiving end and can be divided into pseudo-random codes with a period of 16 and a period of 64, namely a short preamble and a long preamble. The second part is a SIGNAL part, which is used for the receiving end to further acquire necessary information in the SIGNAL transmission process after the short preamble and the long preamble are successfully processed, such as a modulation mode, the number of OFDM packets in a frame, the transmission rate and the like. The third part is the processed data units to be transmitted. The gesture recognition part mainly utilizes the long preamble part as a gesture carrier wave, so that the requirements required by the invention are met on the premise of not influencing WiFi data transmission.

Before the gesture recognition training is carried out, a background noise acquisition process is firstly required. On the premise that no person moves, the environmental signals are collected for a certain time, so that background noise in the later gesture training and recognition process is obtained. And then the signal-to-noise ratio can be greatly improved, and the gesture recognition accuracy is improved to a certain extent. This is because the large size of ambient walls, etc. has a significant advantage in reflecting signal power compared to small area gestures, and the ambient noise changes very little. Therefore, a background noise reduction method can be adopted to obtain the environmental noise under the condition of no target signal and reduce the environmental noise when the target is identified, so that the gesture signal which is not annihilated is obtained.

When the WiFi signal 11 is collected 30 by the receiver and conditioned 31, 32 is transmitted to the MCU data processing chip 4 for signal processing. First, the signal is truncated 40, the long preamble is stripped from the signal and spliced 41 one after the other in order to form a set of gesture signals consisting entirely of long preamble portions. And then, performing down-sampling processing on the signal, taking a window according to a certain length, and taking the average power of the signal in the window as a new signal point to obtain a low-dimensional smooth signal. Thirdly, the signal is subjected to time-frequency transformation processing 43 to obtain a time-frequency matrix thereof and line spectrum extraction 44 is performed on the matrix to obtain a characteristic curve, so that the gesture is recognized.

The gestures are divided into a start gesture, a stop gesture, and a data gesture. The starting gesture marks the start of the operation, the ending gesture marks the end of the gesture operation, and the receiver does not process data any more. The data gestures are a group of gestures, and all the gestures have data corresponding to the gestures one by one for selecting areas and houses, and the areas and the houses are stored in a control matrix according to the gestures and corresponding sequences, so that the electric appliances to be controlled and the changes of the electric appliances to be controlled are obtained.

After the initial gesture is received, the identification of each gesture is completed, the corresponding gesture number is stored in the control matrix, and the operation is repeated until the termination gesture is received. And performing home control according to the control matrix.

As shown in fig. 3, a classification flow chart is shown. And establishing a support vector machine classification model, comparing and identifying the features of the gesture spectral lines, and storing the corresponding bit of the control matrix as 1 when the similarity between the features and the gesture 1 exceeds a threshold. Otherwise, continuing to compare with the gesture 2 until a gesture b with the similarity exceeding the threshold is found, storing the corresponding bit of the control matrix as b, and starting to recognize the next group of gestures.

The control module is schematically shown in fig. 4, and calculates the matrix bit by bit. The regional room 500, the home 501, the parameters to be adjusted (such as the on/off of the lamp, the brightness of the desk lamp and the adjustment 502 of the volume program of the television) and the adjustment modes of the corresponding parameters are selected in sequence, and the appropriate control mode 510, 511 or 512 is selected for control.

In the gesture learning and recognition process, because the gesture changes and has uncontrollable factors such as speed, a dynamic time warping algorithm is adopted. The basic idea is as shown in fig. 5, when the signal trend is constant but the variation speed has a certain difference, the dynamic time warping algorithm is used to warp the time axis, so as to improve the similarity between two groups of signals and reduce the occurrence of missing judgment as much as possible.

The design scheme of the invention can be used for indoor furniture intelligent control, factory control room management and control, shop alarm systems and the like. By the scheme, required operations can be completed quickly and silently.

The invention can provide a quick, effective and low-cost indoor home control system. This system wants the contrast with original intelligent home systems, because the gesture is more clear and definite and the rate of accuracy is higher than pronunciation, because it is better to divide the regional expansibility of setting for to the house, and does not need sound and light based on wiFi's gesture recognition, and is minimum to indoor all the other members' influence, and simultaneously, the settlement of gesture can increase the security, and it takes place to reduce the mistake and touch the condition.

While the invention has been described with reference to the drawings, it is not intended to be limited to the embodiments shown, but rather, it will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention as defined by the appended claims.

In summary, the following steps: the invention designs an intelligent home management system based on a WiFi gesture recognition technology, which comprises the gesture recognition technology under indoor WiFi and the intelligent home management system design; the WiFi-based gesture recognition technology comprises WiFi protocol research, a gesture recognition algorithm and classification model establishment of a support vector machine; the intelligent home management system comprises region selection, home selection, adjustment parameter selection and adjustment state selection. The design finishes gesture recognition and carries out household management control by intercepting the long preamble of the WiFi signal and utilizing time spectrum analysis, line spectrum extraction, support vector machine classification and dynamic time warping algorithm. Therefore, the furniture is rapidly and silently controlled. Avoid the influence that speech control and cell-phone APP control caused.