CN109063697A - A kind of human body sitting posture detection method based on channel state information - Google Patents
A kind of human body sitting posture detection method based on channel state information Download PDFInfo
- Publication number
- CN109063697A CN109063697A CN201811182290.0A CN201811182290A CN109063697A CN 109063697 A CN109063697 A CN 109063697A CN 201811182290 A CN201811182290 A CN 201811182290A CN 109063697 A CN109063697 A CN 109063697A
- Authority
- CN
- China
- Prior art keywords
- sitting posture
- human body
- data
- body sitting
- classification
- 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
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
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/103—Static body considered as a whole, e.g. static pedestrian or occupant recognition
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/30—Noise filtering
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Multimedia (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Human Computer Interaction (AREA)
- Life Sciences & Earth Sciences (AREA)
- Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)
Abstract
A kind of human body sitting posture detection method based on channel state information, it is divided into two stages: off-line training step and on-line testing stage, wherein, complex network is constructed by the analogy of subcarrier sequence signal at time series signal when off-line training step and by visual drawing method to extract network characterization, corresponding statistical nature is extracted from identical data simultaneously, and is combined with two category features and sets up sitting posture fingerprint base；In the on-line testing stage, we carry out the judgement of sitting posture classification and Detection using several machine learning algorithms.We have been combined amplitude and phase information at the same time, and combine multipair antenna further to promote the Stability and veracity of sitting posture detection judgement.Pass through the above method, the present invention can substantially reduce actual operation cost, and overcome the problems, such as that computer vision methods are brought and defect, judgement is detected to the sitting posture of indoor human body to effectively realize, and can optimally realize 95% or more classification accuracy.
Description
Technical field
The present invention relates to human body sitting posture detection field more particularly to a kind of human body sitting posture detections based on channel state information
Method.
Background technique
Seating and standing posture has become a kind of generally existing mode of people in Modern Live, and white collar of going to work is daily in desk
Study that preceding operation computer completes work, student race bends before desk, driver are sitting in driver's cabin manipulation automobile etc. for a long time.So
And sitting for a long time can bring huge negative effect to the body & mind state of people, especially some undesirable sitting postures are practised
It is used, hypometropia is not only easily caused, also easily causes people to suffer from waist, cervical spondylosis and muscle rigidity strain etc. various chronic
Disease.If not prevented and being corrected well, health can be damaged and finally influence personal study and work efficiency.
In addition, some researches show that, the sitting postures of people, there are certain to be associated with his study and work state, therefore, effectively detects and counts people
Sitting posture data, for analyze a people study and work efficiency be far-reaching.With WLAN (WLAN)
Development, hotspot is widely distributed in the places such as various indoor scenarios, such as school, hospital, library, if energy
The wireless device for making full use of these existing realizes human body sitting posture detection judgement, then the cost that system deployment will be substantially reduced,
And the incorrect sitting-pose of energy timely correction people, bring happiness.
Some human body sitting posture detection methods based on computer vision are had already appeared at present, but they mostly exist
Problem and defect.Such as it is traditional based on computer vision methods to illumination require high, the nothing in the case where lacking light
Method realizes accurate judgement.Computer vision methods the problems such as there is also individual privacy and high operation costs.Wireless local area
Net but can preferably overcome the above problem and defect to cause computer vision methods bad in the upper performance of human body sitting posture detection
Problem.
In recent years, WLAN field human body attitude detection, action recognition and in terms of obtain
Many research achievements, and the research of human body sitting posture context of detection is paid close attention to relatively fewer.
Visual Graph building complex network is the research field risen in recent years, and many scholars lead in Visual Graph network
Domain achieves a large amount of research achievement.There are many scholars to have carried out phase to time series signal as tool using Visual Graph network
Close research work.It can be seen that Visual Graph building network is the highly effective tool of processing sequence signal.Of the invention one
Human body sitting posture method of the kind based on channel state information cleverly believes channel state information statistical data feature and channel status
The method of breath Visual Graph is converted into network integration and connects, and realizes human body sitting posture detection judgement.
Summary of the invention
In order to overcome present in the human body sitting posture detection method of active computer vision it is high to ambient lighting requirement, be related to
The deficiency that individual privacy is invaded, operation cost is high, the present invention provide a kind of compare for traditional computer visible sensation method more
Environment is adapted to, is not related to invasion of privacy, has higher accuracy, the more cheap human body sitting posture based on channel state information of cost
Detection method, it can effectively realize that the judgement to indoor human body sitting posture is classified.
In order to solve the above-mentioned technical problem the technical scheme adopted by the invention is that:
A kind of human body sitting posture detection classification method based on channel state information, comprising the following steps:
Step 1: building human body sitting posture channel state information data acquisition platform；
Step 2: experimenter sits quietly in experimental data collection point, human body difference sitting posture is divided into N number of classification, as human body
The basic unit of sitting posture classification, is denoted as L respectively1,…,LN；
Step 3: off-line training step, experimenter keep each different sitting posture for a period of time quiet in data collection point
Only state, so that receiving end acquisition includes the data packet of human body sitting posture channel state information, each channel state information number enough
It is indicated according to packet are as follows:Wherein F1~F30Represent subcarrier, TrFor transmitting antenna number, ReTo receive
Antenna number；
Step 4: take the 1st pair of antenna to pre-process upper amplitude and phase data, process is as follows:
Step 4-1 removes obvious exceptional value particularly pertinent in initial data；
It after step 4-2 is through outlier processing, then with Hampel filter carries out the disposal of gentle filter, it is dry to eliminate noise
It disturbs；
Step 5: first by the analogy of different sub-carrier sequence signal at time series signal, with the Visual Graph of building complex network
Channel state information is converted to network by (Visibility Graph) method, and extracts one from the complex network constructed
A part of feature that a little network characterizations are classified as human body sitting posture；And corresponding amplitude and phase are extracted from same data
Another part feature that statistical nature is also used as human body sitting posture to classify；
Step 6: similarly, since the multipair antenna comprising different human body sitting posture information can be collected into acquisition data procedures
Data can take the amplitude and phase data of different antennae pair, carry out step 4~step 5 repeated work, more to obtain
Validity feature is for distinguishing different human body sitting postures；
Step 7: min-max standardization is carried out to all Network & Stats characteristics for carrying out human body sitting posture classification
Processing, i.e.,Wherein, xnewIndicate normalized treated new feature, xoldIndicate normalized processing
Preceding old feature, xmaxAnd xminRespectively represent the maximum value and minimum value in this feature preceding all samples processed；
Step 8: it will be regarded as a fingerprint deposit sitting posture fingerprint base through all features after min-max standardization,
Training set as human body sitting posture machine learning algorithm is spare；
Step 9: the on-line testing stage, it is identical as the training stage to be again at the progress of identical data collection point by experimenter
Sitting posture experiment and collecting test data packet；
Step 10: data handling procedure identical with the training stage being carried out to test data, i.e., test data is through 4~8 steps
Test set is obtained after rapid processing；
Step 11: human body sitting posture training set spare in sitting posture fingerprint base being called to carry out each sample of test set
Different sitting posture classification, obtain the estimation human body sitting posture of sample under different machines learning algorithm, calculate the standard of different sitting posture classification
True rate.
Further, in the step 5, the process for extracting human body sitting posture characteristic of division is as follows:
Step 5-1: by the amplitude and phase data after processed, principle is constructed by VG:
(1) there must be connection between two neighboring node,
(2) thus the network of method construct be it is nondirectional,
(3) there is connection in the geometrical relationship side for needing to meet in above-mentioned formula between non-conterminous node,
Amplitude or phase are built into complex network in this manner, and thus network extracts 4 network characterizations, it is corresponding
Network characterization respectively degree of being related coefficient, even number of edges, cluster coefficients entropy, weight angle value；
Step 5-2: the amplitude and phase after processed, by mathematical statistics principle:
Mad (X)=median (| Xi-median(X)|)、Median(x)
Above 4 statistical natures of amplitude and phase are extracted respectively.
The beneficial effects of the present invention are:
1. simple WLAN devices, which are utilized, carries out building for human body sitting posture classification stage, conventional machines are overcome
Visible sensation method needs the defect compared with high light conditions, also solves the problems, such as the privacy violation of machine vision method；
2. be detected human body in the present invention without carrying any active equipment, reduce detected person in use by
In the inadaptable sense of bring using active equipment, while further reducing the hardware cost of human body sitting posture detection classification；
3. information status information network and mathematics statistical nature has been used in combination, so that human body sitting posture detection classification is more quasi-
Really, reliably；
4. amplitude and phase information has been used in combination, and adequately the information consolidation of different antennae pair is used, further
Improve the accuracy rate of sitting posture detection.
Detailed description of the invention
Fig. 1 is a kind of distribution schematic diagram of data collection point of the human body sitting posture detection based on channel state information；
Fig. 2 is the method for the present invention and the performance of 30 range values as the original methods of feature is only used to compare figure；
Fig. 3 is confusion matrix schematic diagram of the method for the present invention under optimal SVM algorithm.
Fig. 4 is a kind of flow chart of human body sitting posture detection method based on channel state information.
Specific embodiment
Explanation is described in detail to preferable embodiment of the invention in conjunction with attached drawing in we below, so that advantages of the present invention
It can simpler and be quickly readily appreciated by one skilled in the art with feature, to be made to protection scope of the present invention more clear
Chu explicitly divides and defines.
Referring to Fig.1~Fig. 4, a kind of human body sitting posture detection method based on channel state information, comprising the following steps:
Step 1: building human body sitting posture channel state information data acquisition platform；
Step 2: experimenter sits quietly in experimental data collection point, human body difference sitting posture is divided into N number of classification, as human body
The basic unit of sitting posture classification, is denoted as L respectively1,…,LN；
Step 3: off-line training step, experimenter keep each different sitting posture for a period of time quiet in data collection point
Only state, so that receiving end acquisition includes the data packet of human body sitting posture channel state information, each channel state information number enough
It is indicated according to packet are as follows:Wherein F1~F30It is subcarrier, TrFor transmitting antenna number, ReTo receive day
Line number；
Step 4: take the 1st pair of antenna to carry out pretreatment operation to upper amplitude and phase data, process is as follows:
Step 4-1 removes obvious exceptional value particularly pertinent in initial data；
It after step 4-2 is through outlier processing, then with Hampel filter carries out the disposal of gentle filter, it is dry to eliminate noise
It disturbs；
Step 5: first by the analogy of different sub-carrier signal sequence at time series signal, with the Visual Graph of building complex network
Channel state information is converted to network by (Visibility Graph) method, and extracts one from the complex network constructed
A part of feature that a little network characterizations are classified as human body sitting posture；And corresponding amplitude and phase are extracted from same data
Another part feature that statistical nature is also used as human body sitting posture to classify.Wherein extract the sub-step of network characterization and statistical nature such as
Shown in lower step 5-1 and 5-2.
Step 6: similarly, since the multipair antenna comprising different human body sitting posture information can be collected into acquisition data procedures
Data can take the amplitude and phase data of different antennae pair, carry out the repeated work of step 4~5, more effective to obtain
Feature is for distinguishing different sitting postures；
Step 7: all Network & Stats characteristics for human body sitting posture classification are carried out at min-max standardization
Reason, i.e.,Wherein, xnewIndicate normalized treated new feature, xoldBefore indicating normalized processing
Old feature, xmaxAnd xminRespectively represent the maximum value and minimum value of this feature preceding all samples processed；
Step 8: it will be regarded as a fingerprint deposit sitting posture fingerprint base through all features after min-max standardization,
Training set as human body sitting posture machine learning algorithm is spare；
Step 9: the on-line testing stage, it is identical as the training stage to be again at the progress of identical data collection point by experimenter
Sitting posture experiment and collecting test data packet；
Step 10: data handling procedure identical with the training stage being carried out to test data, i.e., test data is through 4~8 steps
Test set is obtained after rapid processing；
Step 11: human body sitting posture training set spare in sitting posture fingerprint base being called to carry out each sample of test set
Different sitting posture classification, obtain the human body sitting posture estimated result of sample under different machines learning algorithm, calculate different sitting posture classification
Accuracy rate.
Further, in the step 5, two sub-steps for extracting human body sitting posture characteristic of division are made up of:
Step 5-1: by the amplitude and phase data after processed, principle is constructed by VG:
(1) there must be connection between two neighboring node,
(2) thus the network of method construct be it is nondirectional,
(3) there is connection in the geometrical relationship side for needing to meet in above-mentioned formula between non-conterminous node,
Amplitude or phase are built into complex network in this manner, and thus network extracts 4 network characterizations, it is corresponding
Network characterization respectively degree of being related coefficient, even number of edges, cluster coefficients entropy, weight angle value；
Step 5-2: the amplitude and phase after processed, by mathematical statistics principle:
Mad (X)=median (| Xi-median(X)|)、Median(x)
Above 4 statistical natures of amplitude and phase are extracted respectively.
The experiment porch of the present embodiment is made of 2 computers, wherein one is used as access point (AP), another as monitoring
Point (MP) is mounted with 5300 network interface card of Intel wherein being mounted with 5300 network interface card of Intel in AP in MP.Furthermore in order to obtain letter
Channel state information is also mounted with CSI-tool tool for extracting channel state information in MP.
It is student's desk in a classroom that place, which is embodied, and AP and MP will be respectively placed in experimenter two by us
Each 60 centimeters in side, are highly 50 centimetres, and experimenter is located at central seat and shows different sitting postures；
Every time when acquisition, the static data collection point for keeping a sitting posture in such as attached drawing 1 of human body, collecting includes channel shape
The data packet of state information, the acquisition time of each sitting posture are 5 seconds.After acquisition, each difference sitting posture can obtain one
.dat file；
The original channel status information data that each pair of antenna is extracted from the .dat file of each sitting posture, including amplitude
And phase data；
Data are pre-processed, including following two step:
4-1 whether there is exceptional value using 3 times of standard deviation method detection datas, if data are other than 3 times of standard deviation ranges
Then think that it is exceptional value, is then gone to replace exceptional value with sample mean；
4-2 is filtered smooth signal to the data after outlier processing using hampel filter, removal due to
The fluctuation of abnormal signal caused by ambient noise.
Amplitude and phase information after will be preprocessed constructs complex network with the method that VG constructs network, and extracts net
Network feature；Same data are extracted into statistical nature with mathematical statistics principle.The method of two kinds of extraction machine learning classification features
It is made of respectively I and II.
I VG constructs principle:
II mathematical statistics principle:
The statistical natures such as extraction standard difference Std, median absolute deviation Mad, mean value Mean and median Median.
Min-max standardization is carried out to all characteristic classification datas, initial characteristic data passes through standardization
Afterwards, each characteristic value will not influence classification results because certain character numerical values are especially big, be more suitable for carrying out between 0-1
Comprehensive Correlation evaluation；
A fingerprint is regarded as by the feature after min-max standardization, then it is deposited into human body sitting posture and refers to
Line library, it is spare for the training set as machine learning classification；
Test phase, human body equally keep different sitting postures to be located at the test point in Fig. 1, collecting test data, each position
Acquisition time is 5 seconds；
It is special that collected human body sitting posture test data is got into test after operation 4~7 identical with the training stage
Matrix is levied, the resulting test set of data for calling the spare training set in fingerprint base to adopt test phase carries out sitting posture point
Class prediction.
To verify the impact of performance of the invention；We are former with 30 sub- carrier amplitudes using method proposed by the invention and only
The method that initial value makees feature is compared, and it is as shown in Figure 2 to obtain performance.We also depict obscuring for optimal machine learning algorithm
Matrix schematic diagram, as shown in Figure 3.Finally, method proposed by the invention is better than me under two kinds of machine in normal service learning algorithms
Selected control methods, the results showed that method of the invention significantly improves the accuracy rate of classification.In addition, confusion matrix
Good behaviour illustrates that the method for the present invention can distinguish human body difference sitting posture well.To sum up, this method has preferably in an experiment
Expression effect.
The above description is only an embodiment of the present invention, is not intended to limit the scope of the invention, all to utilize this hair
Equivalent structure or equivalent flow shift made by bright specification and accompanying drawing content is applied directly or indirectly in other relevant skills
Art field, is included within the scope of the present invention.
Claims (2)
1. a kind of human body sitting posture based on channel state information detects classification method, which is characterized in that the method includes following
Step:
Step 1: building human body sitting posture channel state information data acquisition platform；
Step 2: experimenter sits quietly in experimental data collection point, human body difference sitting posture is divided into N number of classification, as human body sitting posture
The basic unit of classification, is denoted as L respectively1,…,LN；
Step 3: off-line training step, experimenter keep each the static shape of different sitting postures for a period of time in data collection point
State, so that receiving end acquisition includes the data packet of human body sitting posture channel state information, each channel state information data packet enough
It indicates are as follows:Wherein F1~F30Represent subcarrier, TrFor transmitting antenna number, ReFor receiving antenna
Number；
Step 4: take the 1st pair of antenna to pre-process upper amplitude and phase data, process is as follows:
Step 4-1 removes obvious exceptional value particularly pertinent in initial data；
It after step 4-2 is through outlier processing, then with Hampel filter carries out the disposal of gentle filter, eliminates noise jamming；
Step 5: first by the analogy of different sub-carrier sequence signal at time series signal, with the visual drawing method of building complex network
Channel state information is converted to network, and extracts some network characterizations as human body sitting posture from the complex network constructed
A part of feature of classification；And corresponding amplitude and phase statistical nature is extracted from same data and is also used as human body sitting posture
Another part feature of classification；
Step 6: similarly, since the multipair antenna number comprising different human body sitting posture information can be collected into acquisition data procedures
According to can take the amplitude and phase data of different antennae pair, carry out step 4~step 5 repeated work, more have to obtain
Effect feature is for distinguishing different human body sitting postures；
Step 7: all Network & Stats characteristics for carrying out human body sitting posture classification are carried out at min-max standardization
Reason, i.e.,Wherein, xnewIndicate normalized treated new feature, xoldBefore indicating normalized processing
Old feature, xmaxAnd xminRespectively represent the maximum value and minimum value in this feature preceding all samples processed；
Step 8: it will be regarded as a fingerprint deposit sitting posture fingerprint base through all features after min-max standardization, as
The training set of human body sitting posture machine learning algorithm is spare；
Step 9: the on-line testing stage is again at identical data collection point by experimenter and carries out seat identical with the training stage
Appearance experiment and collecting test data packet；
Step 10: identical with training stage data handling procedure carried out to test data, i.e., test data through 4~8 steps at
Test set is obtained after reason；
Step 11: calling human body sitting posture training set spare in sitting posture fingerprint base to carry out each sample of test set different
Sitting posture classification, obtains the estimation human body sitting posture of sample under different machines learning algorithm, calculates the accuracy rate of different sitting posture classification.
2. a kind of human body sitting posture based on channel state information as described in claim 1 detects classification method, which is characterized in that
In the step 5, the process for extracting human body sitting posture characteristic of division is as follows:
Step 5-1: by the amplitude and phase data after processed, principle is constructed by VG:
(1) there must be connection between two neighboring node,
(2) thus the network of method construct be it is nondirectional,
(3) there is connection in the geometrical relationship side for needing to meet in above-mentioned formula between non-conterminous node,
Amplitude or phase are built into complex network in this manner, and thus network extracts 4 network characterizations, corresponding network
Feature respectively degree of being related coefficient, even number of edges, cluster coefficients entropy, weight angle value；
Step 5-2: the amplitude and phase after processed, by mathematical statistics principle:
Mad (X)=median (| Xi-median(X)|)、Median(x)
Above 4 statistical natures of amplitude and phase are extracted respectively.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811182290.0A CN109063697A (en) | 2018-10-11 | 2018-10-11 | A kind of human body sitting posture detection method based on channel state information |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811182290.0A CN109063697A (en) | 2018-10-11 | 2018-10-11 | A kind of human body sitting posture detection method based on channel state information |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109063697A true CN109063697A (en) | 2018-12-21 |
Family
ID=64763832
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811182290.0A Pending CN109063697A (en) | 2018-10-11 | 2018-10-11 | A kind of human body sitting posture detection method based on channel state information |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109063697A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110062379A (en) * | 2019-04-15 | 2019-07-26 | 哈尔滨工程大学 | Identity identifying method based on channel state information under a kind of human body behavior scene |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106685590A (en) * | 2016-12-08 | 2017-05-17 | 浙江工业大学 | Channel state information and KNN (K-Nearest Neighbor)-based indoor human body orientation recognition method |
| CN107169453A (en) * | 2017-05-16 | 2017-09-15 | 湖南巨汇科技发展有限公司 | A kind of sitting posture detecting method based on depth transducer |
| CN108200644A (en) * | 2018-01-10 | 2018-06-22 | 浙江工业大学 | A kind of passive indoor orientation method based on network Visual Graph |
| CN108268894A (en) * | 2018-01-10 | 2018-07-10 | 浙江工业大学 | A kind of human body based on network Visual Graph is towards detection method |
| CN108549834A (en) * | 2018-03-08 | 2018-09-18 | 佛山市顺德区中山大学研究院 | A kind of human body sitting posture recognition methods and its system based on flexible sensor |
-
2018
- 2018-10-11 CN CN201811182290.0A patent/CN109063697A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106685590A (en) * | 2016-12-08 | 2017-05-17 | 浙江工业大学 | Channel state information and KNN (K-Nearest Neighbor)-based indoor human body orientation recognition method |
| CN107169453A (en) * | 2017-05-16 | 2017-09-15 | 湖南巨汇科技发展有限公司 | A kind of sitting posture detecting method based on depth transducer |
| CN108200644A (en) * | 2018-01-10 | 2018-06-22 | 浙江工业大学 | A kind of passive indoor orientation method based on network Visual Graph |
| CN108268894A (en) * | 2018-01-10 | 2018-07-10 | 浙江工业大学 | A kind of human body based on network Visual Graph is towards detection method |
| CN108549834A (en) * | 2018-03-08 | 2018-09-18 | 佛山市顺德区中山大学研究院 | A kind of human body sitting posture recognition methods and its system based on flexible sensor |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110062379A (en) * | 2019-04-15 | 2019-07-26 | 哈尔滨工程大学 | Identity identifying method based on channel state information under a kind of human body behavior scene |
| CN110062379B (en) * | 2019-04-15 | 2022-10-28 | 哈尔滨工程大学 | Identity authentication method based on channel state information under human behavior scene |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109522793B (en) | Method for detecting and identifying abnormal behaviors of multiple persons based on machine vision | |
| CN107330465B (en) | A kind of images steganalysis method and device | |
| CN106407905B (en) | Wireless aware action identification method based on machine learning | |
| CN101923652B (en) | Pornographic picture identification method based on joint detection of skin colors and featured body parts | |
| CN104143079B (en) | The method and system of face character identification | |
| CN103942577B (en) | Based on the personal identification method for establishing sample database and composite character certainly in video monitoring | |
| CN106980852B (en) | Based on Corner Detection and the medicine identifying system matched and its recognition methods | |
| CN101710383B (en) | Method and device for identity authentication | |
| CN109902558B (en) | CNN-LSTM-based human health deep learning prediction method | |
| CN103793690B (en) | A kind of human-body biological biopsy method detected based on subcutaneous haematic flow and application | |
| CN107491740A (en) | A kind of neonatal pain recognition methods based on facial expression analysis | |
| CN108305187A (en) | Dining room based on computer vision is had dinner guidance system | |
| CN105956570B (en) | Smiling face's recognition methods based on lip feature and deep learning | |
| CN110781953B (en) | Lung cancer pathological section classification method based on multi-scale pyramid convolution neural network | |
| CN106980815A (en) | Facial paralysis objective evaluation method under being supervised based on H B rank scores | |
| CN109241814A (en) | Pedestrian detection method based on YOLO neural network | |
| CN106250701B (en) | The system and method for Lung neoplasm degree of danger classification are realized in computer software | |
| CN109389041A (en) | A kind of fall detection method based on joint point feature | |
| CN108670245A (en) | A kind of electrocardiograph signal detection method and system | |
| CN108710822A (en) | Personnel falling detection system based on infrared array sensor | |
| CN105574509A (en) | Face identification system playback attack detection method and application based on illumination | |
| CN106529377A (en) | Age estimating method, age estimating device and age estimating system based on image | |
| CN106874825A (en) | The training method of Face datection, detection method and device | |
| CN110163157A (en) | A method of more people's Attitude estimations are carried out using novel loss function | |
| CN103034840B (en) | A kind of gender identification method |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20181221 |