CN112600630A - Action identification method and device based on Wi-Fi signal - Google Patents
Action identification method and device based on Wi-Fi signal Download PDFInfo
- Publication number
- CN112600630A CN112600630A CN202011420917.9A CN202011420917A CN112600630A CN 112600630 A CN112600630 A CN 112600630A CN 202011420917 A CN202011420917 A CN 202011420917A CN 112600630 A CN112600630 A CN 112600630A
- Authority
- CN
- China
- Prior art keywords
- signal
- domain information
- action
- frequency
- user
- 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.)
- Pending
Links
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04B—TRANSMISSION
- H04B17/00—Monitoring; Testing
- H04B17/30—Monitoring; Testing of propagation channels
- H04B17/309—Measuring or estimating channel quality parameters
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W24/00—Supervisory, monitoring or testing arrangements
- H04W24/08—Testing, supervising or monitoring using real traffic
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Software Systems (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Data Mining & Analysis (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Medical Informatics (AREA)
- Electromagnetism (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Quality & Reliability (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a Wi-Fi signal-based action recognition method, which comprises the following steps: acquiring a Wi-Fi signal transmitted by a Wi-Fi signal transmitter, and preprocessing the Wi-Fi signal to obtain channel state information of the Wi-Fi signal; carrying out noise reduction processing on the channel state information to obtain at least one group of stable amplitude ratio and phase difference values; filtering the amplitude ratio and the phase difference value, and extracting the characteristics of the amplitude ratio and the phase difference value in the subcarrier with larger difference after filtering; wherein the feature extraction comprises: extracting time domain information and frequency domain information in different time periods; and training and predicting the time domain information and the frequency domain information through a preset machine learning model, and judging and identifying the action of the user. By the method and the device, a user can recognize the action only in a Wi-Fi environment without wearing additional hardware equipment for somatosensory recognition, so that the user has better action recognition experience.
Description
Technical Field
The invention relates to the technical field of Wi-Fi sensing, in particular to a method and a device for identifying actions based on Wi-Fi signals.
Background
Currently, there are three main ways of motion recognition: structured light based devices, camera based and external based devices. The method based on the structured light equipment consists of a set of projector and camera, and recovers a three-dimensional model of a user at each time point by projecting the user and collecting optical signal change caused by user action so as to identify the action, but the structured light system is expensive, and the method is obviously influenced by the environment and is easy to fail to identify due to the change of the environment; the camera-based method collects external images through a monocular camera or a multi-view camera, and realizes action recognition through a machine vision method, but the camera-based method is difficult to use in a dark or dark environment, and the problem of privacy safety is brought when the camera is used for collecting images on household equipment; the external device-based method usually responds to the user action by means of an infrared emitter or a gyroscope and other sensors on the external device to identify the action, but the method needs the user to additionally purchase hardware equipment and correctly wear the hardware equipment, so that the use experience of the user is influenced to a certain extent, and the hardware equipment only responds to the action of the wearing part of the user, so that the integral action is difficult to identify, and the identification effect is often unsatisfactory.
With the widespread use of Wi-Fi devices, some current research has focused on using CSI information of Wi-Fi for motion recognition. Most of the researches pay attention to feature extraction based on time domain and frequency domain or feature identification methods based on machine learning, and the rapid elimination of original signal noise is omitted. The existing research method is difficult to complete the whole action recognition process quickly, one of the important reasons is that a noise elimination module is too complex, a lot of time is needed for noise elimination and feature extraction, a follow-up module is also focused on improving the accuracy of the method, the instantaneity of a real action recognition scene is ignored, and most action recognition use scenes are difficult to meet.
Disclosure of Invention
The purpose of the invention is: the method and the device can control the path through which electromagnetic waves pass by using different antenna pairs, adopt a method for calculating the amplitude ratio and the phase difference between the paths, simply and quickly eliminate most of noises irrelevant to the human body action, extract corresponding characteristics after filtering, and predict the action accuracy by using a machine learning model, thereby achieving the aim of identifying the body action. The whole method does not need a user to wear additional hardware, does not need expensive identification equipment, has high identification speed, and can obtain more characteristic values in a simple mode of increasing the number of antenna pairs, thereby improving the identification accuracy.
In order to achieve the above object, the present invention provides a method for identifying actions based on Wi-Fi signals, comprising: acquiring a Wi-Fi signal transmitted by a Wi-Fi signal transmitter, and preprocessing the Wi-Fi signal to obtain channel state information of the Wi-Fi signal; wherein the Wi-Fi signals comprise Wi-Fi signals which make various corresponding actions by a user; carrying out noise reduction processing on the channel state information to obtain at least one group of stable amplitude ratio and phase difference values; filtering the amplitude ratio and the phase difference value, and extracting the characteristics of the amplitude ratio and the phase difference value in the subcarrier with larger difference after filtering; wherein the feature extraction comprises: extracting time domain information and frequency domain information in different time periods; and training and predicting the time domain information and the frequency domain information through a preset machine learning model, and judging and identifying the action of the user.
Further, the denoising process specifically includes: the amplitude of Wi-Fi signals received by different receiving points at the same time is subjected to ratio processing and the phase is subjected to difference processing.
Further, the time domain information includes: a form factor, a pulse factor, a kurtosis factor, a margin factor, a short-time energy, and a short-time autocorrelation function; the frequency domain information includes: center of gravity frequency, mean square frequency, fundamental frequency, frequency spectrum, energy spectrum, and wavelet coefficients.
Further, the preset machine learning model includes: SVM, GBDT, CNN, and RNN models.
The embodiment of the invention also provides a device for identifying actions based on Wi-Fi signals, which comprises: the device comprises an acquisition module, a data processing module and a judgment module; wherein the content of the first and second substances,
the acquisition module is used for acquiring a Wi-Fi signal transmitted by a Wi-Fi signal transmitter and carrying out preprocessing operation on the Wi-Fi signal to obtain channel state information of the Wi-Fi signal; wherein the Wi-Fi signals comprise Wi-Fi signals which make various corresponding actions by a user;
the data processing module is used for carrying out noise reduction processing on the channel state information to obtain at least one group of stable amplitude ratio and phase difference value; filtering the amplitude ratio and the phase difference value, and extracting the characteristics of the amplitude ratio and the phase difference value in the subcarrier with larger difference after filtering; wherein the feature extraction comprises: extracting time domain information and frequency domain information in different time periods;
and the judging module is used for training and predicting the time domain information and the frequency domain information through a preset machine learning model, and judging and identifying the action of the user.
Further, the denoising process specifically includes: the amplitude of Wi-Fi signals received by different receiving points at the same time is subjected to ratio processing and the phase is subjected to difference processing.
Further, the time domain information includes: a form factor, a pulse factor, a kurtosis factor, a margin factor, a short-time energy, and a short-time autocorrelation function; the frequency domain information includes: center of gravity frequency, mean square frequency, fundamental frequency, frequency spectrum, energy spectrum, and wavelet coefficients.
Further, the preset machine learning model includes: SVM, GBDT, CNN, and RNN models.
An embodiment of the present invention further provides a computer terminal device, which is characterized by including:
one or more processors;
a memory coupled to the processor for storing one or more programs;
when executed by the one or more processors, cause the one or more processors to implement a Wi-Fi signal-based action recognition method as in any above.
An embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored, where the computer program is executed by a processor to implement the method for identifying a Wi-Fi signal-based action according to any one of the above-mentioned items.
Compared with the prior art, the action recognition method and the action recognition device based on the Wi-Fi signals have the advantages that:
1. the user experience is increased: by the method, a user can recognize the action only in a Wi-Fi environment without wearing additional hardware equipment for somatosensory recognition, so that the user has better action recognition experience;
2. and the motion recognition cost is reduced: according to the invention, no external equipment is additionally purchased, expensive equipment such as a camera is not required, only a cheap WI-FI antenna is required, and the hardware cost of action identification can be greatly reduced;
3. can be operated in dark environment: the WI-FI signal, namely the electromagnetic wave signal is adopted for action recognition, so that the method has no requirement on the brightness of the environment, and has low requirement on the use environment compared with a method based on a camera;
4. avoid privacy problems: the invention adopts WI-FI signals to identify actions, compared with the high risk brought to the privacy of the user by camera equipment, the invention only reacts to the specific actions collected in advance, has no identification effect on other actions, has no significance on the signals generated by other actions, and can hardly steal the activity information of the user because the signals are changed greatly after the wall is penetrated.
5. The recognition speed is high: different from other WI-FI action recognition systems, the method simplifies the complex noise reduction process, improves the recognition speed and is more suitable for real-time action recognition scenes.
Drawings
Fig. 1 is a schematic flowchart of a method for identifying a Wi-Fi signal-based action according to a first embodiment of the present invention;
fig. 2 is a schematic overall architecture diagram of a Wi-Fi signal-based action recognition method according to a first embodiment of the present invention;
fig. 3 is a schematic view of a motion sensing game scene in a motion recognition method based on Wi-Fi signals according to an embodiment of the present invention;
fig. 4 is a schematic view of a scene for controlling a smart home in a Wi-Fi signal-based action recognition method according to an embodiment of the present invention;
fig. 5 is a schematic structural diagram of a motion recognition device based on Wi-Fi signals according to a second embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
It should be understood that the step numbers used herein are for convenience of description only and are not intended as limitations on the order in which the steps are performed.
It is to be understood that the terminology used in the description of the invention herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. As used in the specification of the present invention and the appended claims, the singular forms "a," "an," and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise.
The terms "comprises" and "comprising" indicate the presence of the described features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof.
The term "and/or" refers to and includes any and all possible combinations of one or more of the associated listed items.
The first embodiment of the present invention:
as shown in fig. 1 to 4, a method for identifying a motion based on a Wi-Fi signal according to an embodiment of the present invention includes at least the following steps:
s101, acquiring a Wi-Fi signal transmitted by a Wi-Fi signal transmitter, and preprocessing the Wi-Fi signal to obtain channel state information of the Wi-Fi signal; wherein the Wi-Fi signals comprise Wi-Fi signals which make various corresponding actions by a user;
it should be noted that, for step S101, the Wi-Fi signal transmitter and receiver structure and function are as follows:
1.1) transmitting end antenna group: is responsible for the emission of Wi-Fi signals. The antenna group at the transmitting end needs at least two antennas, the more the number of the antennas used in the antenna group is, the better the identification effect is, and each time one antenna is added or reduced, or other antennas at different positions are used, the new antenna array is adopted;
1.2) a signal transmitting module: and the system is responsible for receiving the instruction of the user terminal and generating a specified format of directional data packet to be transmitted. The module is completed by an inner chip of a Wi-Fi signal transmitter, and the Wi-Fi signal transmitter is equipment for transmitting Wi-Fi signals and comprises but is not limited to a Wi-Fi router, a mobile phone, intelligent household equipment, a motion sensing game machine and the like;
1.3) receiving end antenna group: is responsible for the reception of Wi-Fi signals. The receiving end antenna group needs at least one antenna, the more the number of the antennas used in the antenna group is, the better the identification effect is, and each time one antenna is increased or decreased, or other antennas in different positions are used, the new antenna array is adopted;
1.4) a signal receiving module: and the device is responsible for receiving the directional data packet with the specified format sent by the Wi-Fi signal transmitter. The module is completed by an inner chip of a Wi-Fi signal receiver, and the Wi-Fi signal receiver is equipment for receiving Wi-Fi signals and comprises but is not limited to a Wi-Fi router, a mobile phone, intelligent household equipment, a motion sensing game machine and the like.
It should be noted that, for the preprocessing operation, the method includes:
2.1) data packet analysis: the data packet with the specified format obtained by the signal receiving module is analyzed into a channel state information packet;
2.2) extracting amplitude and phase: and processing the channel state information matrix in the channel state information packet into amplitude and phase information of electromagnetic waves, and sending the amplitude and phase information to the noise reduction module.
S102, carrying out noise reduction processing on the channel state information to obtain at least one group of stable amplitude ratio and phase difference values; filtering the amplitude ratio and the phase difference value, and extracting the characteristics of the amplitude ratio and the phase difference value in the subcarrier with larger difference after filtering; wherein the feature extraction comprises: extracting time domain information and frequency domain information in different time periods;
it should be noted that the generation of the amplitude ratio and the phase difference specifically includes: the amplitude and phase information is generated to the required amplitude ratio and phase difference information depending on the antenna set used. The amplitude ratio and the phase difference refer to the amplitude ratio and the phase difference between two or more groups of antenna pairs with different electromagnetic wave paths between transmitting antennas and receiving antennas in the used antenna array, wherein the transmitting and receiving antennas can be regarded as a new group of antenna pairs as long as one antenna is different;
it should be noted that, for the filtering process, the butterworth filter, the high-pass filter or other filters with similar functions of the present invention filter the obtained amplitude ratio and phase difference, retain the fluctuation caused by the human body movement in a certain frequency range, and eliminate the fluctuation caused by other factors besides the human body movement. The frequency range of filtering may be selected based on the identified motion. The amplitude ratio and the phase difference value under different filtered subcarriers are selected, the subcarrier with larger fluctuation of the amplitude ratio and the phase difference value after filtering can be selected as the optimal subcarrier by using methods of selecting the maximum variance and the like, and the corresponding amplitude ratio and the corresponding phase difference are used as action samples to be sent to the characteristic extraction module.
It should be noted that, for feature extraction, specifically: the motion samples are segmented into time segments that can be used for identification. A reference method is given below: according to the difference of the identification action and the difference of the action occurrence time, in the time window needing action identification, an action sample is divided into proper time slices by using a windowing method or other methods, the time slices with small fluctuation are removed, no effective action occurs in the time, the time slices with large fluctuation are reserved, and the time slices with large fluctuation are subdivided to obtain the proper small time slices for identification. The time slices are sequential;
and extracting the time domain characteristics and the frequency domain characteristics of fluctuation in each time slice according to the identified action and the requirement of identification precision. Time domain features include, but are not limited to, form factors, impulse factors, kurtosis factors, margin factors, short-term energies, short-term autocorrelation functions, and the like, and frequency domain features include, but are not limited to, center-of-gravity frequencies, mean-square frequencies, fundamental frequencies, frequency spectra, energy spectra, wavelet coefficients, and the like. The spatial characteristics can also be extracted by training models such as CNN and RNN.
S103, training and predicting the time domain information and the frequency domain information through a preset machine learning model, and judging and identifying the action of the user.
It should be noted that, for the preset machine learning model, this part is performed before the system is built, and the trained model is directly loaded into the system during the system building without repeated training. The model training process is as follows: obtaining a plurality of groups of eigenvalues and corresponding information such as subcarriers, equipment spacing, environment types, used antenna arrays, action standard degrees and the like under standard actions and non-standard actions to form a training data set, and training a prediction model capable of predicting the action standard degrees according to the information by using an SVM (support vector machine), a GBDT (GBDT) or other machine learning models. The prediction model is responsible for predicting the action standard degree according to the characteristics of the action sample and the test environment thereof, and generating a corresponding instruction or feeding back the instruction to a user according to the action standard degree.
In an embodiment of the present invention, the denoising process specifically includes: the amplitude of Wi-Fi signals received by different receiving points at the same time is subjected to ratio processing and the phase is subjected to difference processing.
In one embodiment of the present invention, the time domain information includes: a form factor, a pulse factor, a kurtosis factor, a margin factor, a short-time energy, and a short-time autocorrelation function; the frequency domain information includes: center of gravity frequency, mean square frequency, fundamental frequency, frequency spectrum, energy spectrum, and wavelet coefficients.
In one embodiment of the present invention, the preset machine learning model includes: SVM, GBDT, CNN, and RNN models.
For a better understanding of the present invention, it can be specifically understood from the following examples:
example 1, as shown in fig. 3, is a specific embodiment in which a Wi-Fi signal-based action recognition method is applied to a motion sensing game scene. The Wi-Fi signal transmitting end is a Wi-Fi router, the Wi-Fi signal receiving end is a somatosensory game host, signal receiving, signal processing, noise reduction, feature extraction and model use are completed by the somatosensory game host, the model training stage is completed during game development, a model to be used is loaded into the game host when a game is loaded, and the display is connected with the somatosensory game host and is responsible for displaying pictures output by the somatosensory game host. The user stands at a designated position, and can start the motion sensing game after sending a game starting command to the game host machine by using a gamepad or other modes. At the moment, the user plays a body feeling dance game. The specific process of application is as follows:
1. a user stands at a designated position, sends a game starting instruction to the motion sensing game host through the game handle, and displays corresponding dance guide actions along with the progress of music to the user through the display;
2. when the user makes corresponding action, the WI-FI router sends an identification data packet to the motion sensing game host;
3. a receiving antenna of the motion sensing game host receives a data packet containing information and transmits the data packet to the signal processing module;
4. the signal processing module converts the data packet into channel state information of an electromagnetic wave signal, converts the channel state information into corresponding electromagnetic wave amplitude and phase information, and transmits the information to the noise reduction module;
5. after the noise reduction module generates the amplitude ratio and the phase difference which correspond to the antenna pair, the noise reduction module filters in the selected frequency range by using a high-pass filter, selects the optimal subcarrier after filtering, and transmits the corresponding amplitude ratio and phase difference information to the feature extraction module;
6. the characteristic extraction module carries out time slice segmentation operation on the information, extracts corresponding time domain and frequency domain information and transmits the information to the model training module;
7. the model training module predicts the action accuracy rate through information, scores the actions of the user according to a preset rule, transmits the scores to a display and feeds the scores back to the user;
8. at the moment, the music just goes to the next action, and after the user receives feedback, the user continues to follow the music to make the next dance action, and the steps are repeated.
Example 2, as shown in fig. 4, is a specific embodiment in which a Wi-Fi signal-based action recognition method is applied to an intelligent home scene. The Wi-Fi signal transmitting end is a Wi-Fi router, the Wi-Fi signal receiving end is an intelligent desk lamp connected with the intelligent home system, meanwhile, the intelligent home system is controlled by an intelligent sound box, signal receiving, signal processing, noise reduction, feature extraction and model use are completed by the intelligent sound box, the model training stage is completed when the system is developed, and the model to be used is loaded into the intelligent sound box when the system is loaded. The user is located appointed region, carries out gesture control to intelligent desk lamp. The specific process of application is as follows:
1. the user is positioned in the designated area and makes a V-shaped gesture on the desk lamp;
2. when the user makes corresponding action, the WI-FI router sends an identification data packet to the intelligent desk lamp;
3. a receiving antenna of the intelligent desk lamp receives a data packet containing information and transmits the data packet to a data processing module of the intelligent sound box;
4. the signal processing module converts the data packet into channel state information of an electromagnetic wave signal, converts the channel state information into corresponding electromagnetic wave amplitude and phase information, and transmits the information to the noise reduction module;
5. after the noise reduction module generates the amplitude ratio and the phase difference which correspond to the antenna pair, the noise reduction module filters in the selected frequency range by using a high-pass filter, selects the optimal subcarrier after filtering, and transmits the corresponding amplitude ratio and phase difference information to the feature extraction module;
6. the characteristic extraction module carries out time slice segmentation operation on the information, extracts corresponding time domain and frequency domain information and transmits the information to the model training module;
7. the model training module predicts the action accuracy rate through the information, generates a light-on command according to the accuracy rate and transmits the light-on command to the intelligent desk lamp;
8. the intelligent desk lamp receives the instruction and turns on the lamp.
Compared with the prior art, the action recognition method based on the Wi-Fi signal has the following beneficial effects that:
1. the user experience is increased: by the method, a user can recognize the action only in a Wi-Fi environment without wearing additional hardware equipment for somatosensory recognition, so that the user has better action recognition experience;
2. and the motion recognition cost is reduced: according to the invention, no external equipment is additionally purchased, expensive equipment such as a camera is not required, only a cheap WI-FI antenna is required, and the hardware cost of action identification can be greatly reduced;
3. can be operated in dark environment: the WI-FI signal, namely the electromagnetic wave signal is adopted for action recognition, so that the method has no requirement on the brightness of the environment, and has low requirement on the use environment compared with a method based on a camera;
4. avoid privacy problems: the invention adopts WI-FI signals to identify actions, compared with the high risk brought to the privacy of the user by camera equipment, the invention only reacts to the specific actions collected in advance, has no identification effect on other actions, has no significance on the signals generated by other actions, and can hardly steal the activity information of the user because the signals are changed greatly after the wall is penetrated.
5. The recognition speed is high: different from other WI-FI action recognition systems, the method simplifies the complex noise reduction process, improves the recognition speed and is more suitable for real-time action recognition scenes.
Second embodiment of the invention:
as shown in fig. 5, an embodiment of the present invention provides a device 200 for identifying actions based on Wi-Fi signals, including: an acquisition module 201, a data processing module 202 and a judgment module 203; wherein the content of the first and second substances,
the acquisition module 201 is configured to acquire a Wi-Fi signal transmitted by a Wi-Fi signal transmitter, and perform a preprocessing operation on the Wi-Fi signal to obtain channel state information of the Wi-Fi signal; wherein the Wi-Fi signals comprise Wi-Fi signals which make various corresponding actions by a user;
the data processing module 202 is configured to perform noise reduction processing on the channel state information to obtain at least one group of stable amplitude ratio and phase difference; filtering the amplitude ratio and the phase difference value, and extracting the characteristics of the amplitude ratio and the phase difference value in the subcarrier with larger difference after filtering; wherein the feature extraction comprises: extracting time domain information and frequency domain information in different time periods;
the judging module 203 is configured to train and predict the time domain information and the frequency domain information through a preset machine learning model, and judge and recognize a user action.
In an embodiment of the present invention, the denoising process specifically includes: the amplitude of Wi-Fi signals received by different receiving points at the same time is subjected to ratio processing and the phase is subjected to difference processing.
In one embodiment of the present invention, the time domain information includes: a form factor, a pulse factor, a kurtosis factor, a margin factor, a short-time energy, and a short-time autocorrelation function; the frequency domain information includes: center of gravity frequency, mean square frequency, fundamental frequency, frequency spectrum, energy spectrum, and wavelet coefficients.
In one embodiment of the present invention, the preset machine learning model includes: SVM, GBDT, CNN, and RNN models.
Third embodiment of the invention:
the embodiment of the invention provides computer terminal equipment, which comprises one or more processors; a memory coupled to the processor for storing one or more programs; when executed by the one or more processors, cause the one or more processors to implement a Wi-Fi signal-based action recognition method as in any above.
The fourth embodiment of the present invention:
an embodiment of the present invention provides a computer-readable storage medium, on which a computer program is stored, where the computer program, when executed by a processor, implements the action recognition method based on Wi-Fi signals as described in any one of the above.
In conclusion, according to the invention, the user can recognize the action only in a Wi-Fi environment without wearing additional hardware equipment for body feeling recognition, so that the user has better action recognition experience. The invention has low requirements on the performance and the computing power of the antenna and strong portability, and can be transplanted to any equipment with enough computing power and electromagnetic wave transceiving function.
The above description is only a preferred embodiment of the present invention, and it should be noted that, for those skilled in the art, various modifications and substitutions can be made without departing from the technical principle of the present invention, and these modifications and substitutions should also be regarded as the protection scope of the present invention.
Claims (10)
1. A motion recognition method based on Wi-Fi signals is characterized by comprising the following steps:
acquiring a Wi-Fi signal transmitted by a Wi-Fi signal transmitter, and preprocessing the Wi-Fi signal to obtain channel state information of the Wi-Fi signal; wherein the Wi-Fi signals comprise Wi-Fi signals which make various corresponding actions by a user;
carrying out noise reduction processing on the channel state information to obtain at least one group of stable amplitude ratio and phase difference values; filtering the amplitude ratio and the phase difference value, and extracting the characteristics of the amplitude ratio and the phase difference value in the subcarrier with larger difference after filtering; wherein the feature extraction comprises: extracting time domain information and frequency domain information in different time periods;
and training and predicting the time domain information and the frequency domain information through a preset machine learning model, and judging and identifying the action of the user.
2. The method for motion recognition based on Wi-Fi signals according to claim 1, wherein the denoising process specifically comprises: the amplitude of Wi-Fi signals received by different receiving points at the same time is subjected to ratio processing and the phase is subjected to difference processing.
3. The Wi-Fi signal-based action recognition method of claim 1, wherein the time domain information comprises: a form factor, a pulse factor, a kurtosis factor, a margin factor, a short-time energy, and a short-time autocorrelation function; the frequency domain information includes: center of gravity frequency, mean square frequency, fundamental frequency, frequency spectrum, energy spectrum, and wavelet coefficients.
4. The Wi-Fi signal-based motion recognition method of claim 1, wherein the preset machine learning model comprises: SVM, GBDT, CNN, and RNN models.
5. A Wi-Fi signal-based motion recognition apparatus, comprising: the device comprises an acquisition module, a data processing module and a judgment module; wherein the content of the first and second substances,
the acquisition module is used for acquiring a Wi-Fi signal transmitted by a Wi-Fi signal transmitter and carrying out preprocessing operation on the Wi-Fi signal to obtain channel state information of the Wi-Fi signal; wherein the Wi-Fi signals comprise Wi-Fi signals which make various corresponding actions by a user;
the data processing module is used for carrying out noise reduction processing on the channel state information to obtain at least one group of stable amplitude ratio and phase difference value; filtering the amplitude ratio and the phase difference value, and extracting the characteristics of the amplitude ratio and the phase difference value in the subcarrier with larger difference after filtering; wherein the feature extraction comprises: extracting time domain information and frequency domain information in different time periods;
and the judging module is used for training and predicting the time domain information and the frequency domain information through a preset machine learning model, and judging and identifying the action of the user.
6. The Wi-Fi signal-based motion recognition device according to claim 5, wherein the denoising process is specifically: the amplitude of Wi-Fi signals received by different receiving points at the same time is subjected to ratio processing and the phase is subjected to difference processing.
7. The Wi-Fi signal-based action recognition device of claim 5, wherein the time domain information comprises: a form factor, a pulse factor, a kurtosis factor, a margin factor, a short-time energy, and a short-time autocorrelation function; the frequency domain information includes: center of gravity frequency, mean square frequency, fundamental frequency, frequency spectrum, energy spectrum, and wavelet coefficients.
8. The Wi-Fi signal-based motion recognition device of claim 5, wherein the preset machine learning model comprises: SVM, GBDT, CNN, and RNN models.
9. A computer terminal device, comprising:
one or more processors;
a memory coupled to the processor for storing one or more programs;
the one or more programs, when executed by the one or more processors, cause the one or more processors to implement the Wi-Fi signal-based action recognition method of any of claims 1-4.
10. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, implements the Wi-Fi signal-based action recognition method according to any one of claims 1 to 4.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011420917.9A CN112600630A (en) | 2020-12-07 | 2020-12-07 | Action identification method and device based on Wi-Fi signal |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011420917.9A CN112600630A (en) | 2020-12-07 | 2020-12-07 | Action identification method and device based on Wi-Fi signal |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112600630A true CN112600630A (en) | 2021-04-02 |
Family
ID=75188686
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011420917.9A Pending CN112600630A (en) | 2020-12-07 | 2020-12-07 | Action identification method and device based on Wi-Fi signal |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112600630A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2023222087A1 (en) * | 2022-05-20 | 2023-11-23 | 华为技术有限公司 | Sensing method and apparatus |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107968689A (en) * | 2017-12-06 | 2018-04-27 | 北京邮电大学 | Perception recognition methods and device based on wireless communication signals |
| CN107994960A (en) * | 2017-11-06 | 2018-05-04 | 北京大学（天津滨海）新代信息技术研究院 | A kind of indoor activity detection method and system |
| CN109171731A (en) * | 2018-09-04 | 2019-01-11 | 北京大学（天津滨海）新代信息技术研究院 | A kind of contactless breathing detection method |
| CN109416294A (en) * | 2016-07-08 | 2019-03-01 | 帝国创新有限公司 | Device and method for generating the movement signature of the movement of the moving component of instruction target machine |
| CN110443206A (en) * | 2019-08-07 | 2019-11-12 | 北京邮电大学 | A kind of human body attitude image generating method and device based on Wi-Fi signal |
| CN112036433A (en) * | 2020-07-10 | 2020-12-04 | 天津城建大学 | CNN-based Wi-Move behavior sensing method |
-
2020
- 2020-12-07 CN CN202011420917.9A patent/CN112600630A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109416294A (en) * | 2016-07-08 | 2019-03-01 | 帝国创新有限公司 | Device and method for generating the movement signature of the movement of the moving component of instruction target machine |
| CN107994960A (en) * | 2017-11-06 | 2018-05-04 | 北京大学（天津滨海）新代信息技术研究院 | A kind of indoor activity detection method and system |
| CN107968689A (en) * | 2017-12-06 | 2018-04-27 | 北京邮电大学 | Perception recognition methods and device based on wireless communication signals |
| US20190174330A1 (en) * | 2017-12-06 | 2019-06-06 | Beijing University Of Posts & Telecommunications | Sensing recognition method and device based on wireless communication signals |
| CN109171731A (en) * | 2018-09-04 | 2019-01-11 | 北京大学（天津滨海）新代信息技术研究院 | A kind of contactless breathing detection method |
| CN110443206A (en) * | 2019-08-07 | 2019-11-12 | 北京邮电大学 | A kind of human body attitude image generating method and device based on Wi-Fi signal |
| CN112036433A (en) * | 2020-07-10 | 2020-12-04 | 天津城建大学 | CNN-based Wi-Move behavior sensing method |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2023222087A1 (en) * | 2022-05-20 | 2023-11-23 | 华为技术有限公司 | Sensing method and apparatus |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107968689B (en) | Perception identification method and device based on wireless communication signals | |
| US10453457B2 (en) | Method for performing voice control on device with microphone array, and device thereof | |
| EP3923273A1 (en) | Voice recognition method and device, storage medium, and air conditioner | |
| CN103353935A (en) | 3D dynamic gesture identification method for intelligent home system | |
| CN109448705B (en) | Voice segmentation method and device, computer device and readable storage medium | |
| CN103413113A (en) | Intelligent emotional interaction method for service robot | |
| CN111124108B (en) | Model training method, gesture control method, device, medium and electronic equipment | |
| CN107066081B (en) | Interactive control method and device of virtual reality system and virtual reality equipment | |
| CN110852200B (en) | Non-contact human body action detection method | |
| CN107004426A (en) | The method and mobile terminal of the sound of admission video recording object | |
| CN111225237B (en) | Sound and picture matching method of video, related device and storage medium | |
| CN105446362B (en) | Posture detection based on computer science auxiliary adjusts devices and methods therefor | |
| CA2979271A1 (en) | Wayfinding and obstacle avoidance system | |
| CN111930336A (en) | Volume adjusting method and device of audio device and storage medium | |
| CN109186752B (en) | Underwater acoustic signal acquisition, transmission and detection system based on graphic processor | |
| CN112600630A (en) | Action identification method and device based on Wi-Fi signal | |
| CN107452381B (en) | Multimedia voice recognition device and method | |
| CN115343704A (en) | Gesture recognition method of FMCW millimeter wave radar based on multi-task learning | |
| CN106507201A (en) | A kind of video playing control method and device | |
| CN112420073A (en) | Voice signal processing method, device, electronic equipment and storage medium | |
| CN113406610B (en) | Target detection method, device, equipment and storage medium | |
| CN106610716A (en) | Gesture recognition method and device | |
| CN110572600A (en) | video processing method and electronic equipment | |
| CN108260012B (en) | Electronic device, video playing control method and related product | |
| WO2017156021A1 (en) | Object detection, analysis, and alert system for use in providing visual information to the blind |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210402 |
|
| RJ01 | Rejection of invention patent application after publication |