CN106406516A - Local real-time movement trajectory characteristic extraction and identification method for smartphone - Google Patents
Local real-time movement trajectory characteristic extraction and identification method for smartphone Download PDFInfo
- Publication number
- CN106406516A CN106406516A CN201610732089.XA CN201610732089A CN106406516A CN 106406516 A CN106406516 A CN 106406516A CN 201610732089 A CN201610732089 A CN 201610732089A CN 106406516 A CN106406516 A CN 106406516A
- Authority
- CN
- China
- Prior art keywords
- mobile phone
- data
- user
- smartphone
- smart mobile
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
- G06F3/01—Input arrangements or combined input and output arrangements for interaction between user and computer
- G06F3/017—Gesture based interaction, e.g. based on a set of recognized hand gestures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/08—Feature extraction
- G06F2218/10—Feature extraction by analysing the shape of a waveform, e.g. extracting parameters relating to peaks
Landscapes
- Engineering & Computer Science (AREA)
- General Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Human Computer Interaction (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Mobile Radio Communication Systems (AREA)
- Telephone Function (AREA)
Abstract
The invention relates to a local real-time movement trajectory characteristic extraction and identification method for a smartphone. The method has two stages including a training stage and an identifying stage; in the training stage, a user carries the smartphone to do different actions; data of a triaxial acceleration sensor is acquired; for a user posture behaviour, peak and trough characteristic points are extracted, so that binary coding of the peak and trough characteristic points is carried out; after being coded, the characteristic points are vectorized; furthermore, matrixing of actions having the same peak and trough number value is carried out; multiple samples are acquired and trained on an upper computer; therefore, a user action characteristic standard library is established; in the identifying stage, the user action standard library is transferred to the smartphone; the user carries the smartphone to do actions; sensor data is acquired at the smartphone side; characteristic point extraction of the smartphone is carried out locally; the user action standard library is matched; and thus, user actions are identified. According to the method disclosed by the invention, characteristic vectors needing to be stored by the movement trajectory of the smartphone are identified as a binary system; furthermore, the data size is small; the identification process is relatively simple; massive calculation does not need to be performed; and thus, the method is suitable for embedded equipment, such as the resource limited smartphone.
Description
Technical field
The present invention relates to smart mobile phone track identification technical field, refer specifically to smart mobile phone locally real-time movement locus
Feature extraction and recognition method.
Background technology
Embedded in mobile phone has acceleration transducer, can be used to detect size and the side of mobile phone acceleration according to acceleration transducer
To, and then perceive the state in three dimensions.There are X, the coordinate in tri- directions of Y, Z for mobile phone acceleration sensor, work as user
When carrying mobile phone does different actions, for X, tri- change in coordinate axis direction of Y, Z all can produce different numerical value, and three axles produce simultaneously
Data reflect the state of user.More complicated for smart mobile phone movement locus Feature extraction and recognition process at present, and
Smart mobile phone computing capability is weak, and memory space is little, needs the movement locus feature carrying out non-real time by extras to carry
The present situation taking and identifying is it is impossible to meet the identification that Smartphone device locally carries out movement locus in real time.
Content of the invention:
The present invention proposes a kind of smart mobile phone locally real-time movement locus Feature extraction and recognition method, and the method identifies mobile phone
For binary system and data scale is little for the characteristic point vector of the required storage of movement locus, and identification process is relatively simple need not to be carried out in a large number
Calculate, be suitable for carrying out on the embedded devices such as resource-constrained smart mobile phone.
For this reason, the technical scheme being adopted is：
A kind of smart mobile phone locally real-time movement locus Feature extraction and recognition method, the method is divided into training and identification two
Stage；In the training stage：User carries smart mobile phone and does different actions, gathers separated intelligent mobile phone 3-axis acceleration sensor
Data；Wave crest and wave trough characteristic point is extracted to user's attitude behavior；Binary coding quantization is carried out to Wave crest and wave trough characteristic point；Coding
Characteristic point vectorization afterwards；And Wave crest and wave trough quantity identical action is carried out matrixing, gather multiple samples and carry out in host computer
Training, sets up user action characteristic standard storehouse；In cognitive phase：Transplanting user action characteristic standard storehouse is to smart mobile phone, user
When carrying smart mobile phone and doing different actions, gather sensing data in mobile phone terminal, open up relief area in mobile phone terminal, extract motion
Track data, sets up multithreading and extracts segmentation feature point, mate user action characteristic standard storehouse, smart mobile phone locally enters in real time
Row feature point extraction, and then identifying user action.
Complicated relative to smart mobile phone movement locus Feature extraction and recognition process, and smart mobile phone computing capability
Weak, memory space is little, needs to carry out the present situation of the movement locus Feature extraction and recognition of non-real time by extras, this
Invention has advantages below to the identification of smart mobile phone movement locus：(1) smart mobile phone can be directed to or some are small-sized embedded
Formula equipment carries out the identification of movement locus.(2) calculate characteristic point simple, reduce the normalized to characteristic point, be suitable for calculating
The weak smart mobile phone of ability.(3) after quantization encoding being carried out to movement locus characteristic point, the required storage of identification mobile phone movement locus
Characteristic point vector be binary system and data scale is little, the embedded device such as smart mobile phone limited by suitable storage resource.
Brief description：
Fig. 1 is that smart mobile phone movement locus of the present invention train flow chart；
Fig. 2 is smart mobile phone movement locus identification process figure of the present invention；
Fig. 3 produces data waveform figure for intelligent mobile phone sensor of the present invention；
Fig. 4 is separation sensor data of the present invention and feature extraction code pattern；
Fig. 5 utilizes windows detecting Wave crest and wave trough figure for the present invention；
Fig. 6 is Wave crest and wave trough characteristic vector figure of the present invention；
Fig. 7 is special action Wave crest and wave trough eigenmatrix figure of the present invention.
Specific embodiment：
Below in conjunction with the accompanying drawings the present invention is described in further detail.
The training schematic flow sheet of the present invention is as shown in Figure 1：
It is that user carries smart mobile phone or low profile edge equipment does different actions first, 3-axis acceleration sensor produces number
According to according to X, Y, data is carried out separating by Z axis, then finds Wave crest and wave trough characteristic point in all directions, and is encoded, will
Characteristic point vectorization, the same number of attitude matrix for Wave crest and wave trough, gather multiple samples, using artificial neural network instruction
Practice sample, form user action characteristic standard storehouse.
By smart mobile phone acceleration transducer along X, Y, Z tri- axle carries out separating.When having in X-direction for a certain action
It is unique, you can identification.If in X-axis None- identified, for Y, Z axis just need not be compared.And successively to Y-axis, Z axis are carried out
Process.The process in mobile phone end data can be reduced using the method, save smart mobile phone computing resource.
Form using one piece of data segmentation extracts Wave crest and wave trough characteristic point to user's attitude behavior.This method provide one
Plant the extracting mode of data Wave crest and wave trough characteristic point, method is simple, amount of calculation is less and can filter miscellaneous point and find out crucial spy
Levy a little.
The form of the Wave crest and wave trough characteristic point vector after coding is represented, can be by feature dot format by vectorization
Change, conveniently calculate further, and computational efficiency can be improved.
The same number of action of Wave crest and wave trough carries out matrixing process.The same number of action of Wave crest and wave trough forms oneMatrix, M represents the number of Wave crest and wave trough on three coordinate axess.Further increase computational efficiency.
The identification process schematic diagram of the present invention is as shown in Figure 2：
First user action characteristic standard storehouse is transplanted in smart mobile phone memory space.User carry smart mobile phone do different
During action, gather sensing data in mobile phone terminal, open up relief area in mobile phone terminal, extract motion trace data, set up multithreading
Extract segmentation feature point, mate user action characteristic standard storehouse, smart mobile phone locally carries out feature point extraction, Jin Ershi in real time
Other user action.
Flow process will be further described below.
As shown in Figure 3：When user does action, smart mobile phone acceleration transducer produces data in X, tri- directions of Y, Z,
Data is processed, then splicing is it is simply that three waveforms.Data on mobile phone coordinate axess reflects the kinestate of user.
As shown in Figure 4：According to intelligent mobile phone sensor X, three groups of device data difference to acceleration sensing for tri- directions of Y, Z
Processed, the order according to Wave crest and wave trough appearance is as the characteristic point of waveform.It is 1 according to crest, trough is that 0 mode will be examined
The characteristic point measuring is encoded.Computer Storage resource can be saved using binary coding.
The method of Wave crest and wave trough feature extraction：
As shown in Figure 5：Identify crest and the trough of one piece of data using sliding window.Data by the one section of fixed number producing
As definite value sliding window, user movement status data all, in definite value sliding window, sliding window is divided into equal portions
Little sliding window, carries out the detection of Wave crest and wave trough data in this little sliding window, can have in each little sliding window crest with
Trough, the order according to occurring is ranked up, by each comparing of crest and trough, front several crests of value maximum and value
Minimum rear several troughs, as the characteristic point of each axle.Partly miscellaneous point can be filtered out by this method, search out key
Characteristic point.
As shown in Figure 6：The characteristic point that Wave crest and wave trough is obtained carries out vectorization.The data that three groups of acceleration transducers produce
The Wave crest and wave trough producing sequentially in time, forms three groups of vectors, and the data in each vector represents the characteristic point of all directions
And the order that characteristic point occurs.
As shown in Figure 7：The data that special action is formed carries out matrixing process.Separation sensor all directions produce
Data, if three coordinate axess Wave crest and wave trough length are all identical, can carry out matrixing process.Form oneMatrix.
Cognitive phase：Each moment of intelligent mobile phone sensor can produce data, for data Wave crest and wave trough calculating with
The generation of data has synchronous problem, and using developing relief area, and multithreading solves stationary problem.
Implementation method is as follows：During Wave crest and wave trough detection, define a big sliding window relief area, will sense
The continuous iteration of data that device produces is deposited into oneIn fixed length matrix, wherein N set according to display windows length one
Individual length value, adds latest data in matrix, abandons the relatively early data being stored in simultaneously, and the length maintaining matrix is fixed value N,
Then by 3 fixed length matrix deciles, the little sliding window relief area after decile is opened up with multiple threads and carries out Wave crest and wave trough spy
The detection levied.
In user movement state with Sample Storehouse comparison process, first X-axis data is compared, if for a certain action
There is it unique in X-direction, you can identification.If in X-axis None- identified, for Y, Z axis just need not be compared.Successively to Y
Axle, Z axis are processed.
Further illustrate the effect of the present invention below by specific application scenarios：
Scene 1：Field of human-computer interaction.Present man-machine interaction is required for buying supporting equipment, spends costly.Intelligence
Mobile phone arranges many sensors as the very high equipment of an integrated level, inside, all the time all create substantial amounts of
Data, and smart mobile phone computing capability is weak, memory space is little, can carry out effective storage of key point for the data producing.
Identifying user action, and then map other functions, carry out man-machine interaction.
Scene 2：Characteristic point normalized.For some wave character process aspects, it is subject to a lot of disturbing sometimes,
Such as speed, power, the characteristic point that collects is needed to be normalized, by this method may not necessarily carry out with big than
Example or the normalized with small scale, only need to find Wave crest and wave trough characteristic point sequentially in time.
Scene 3：Characteristic point quantifies and calculates.For some specific fields, it is impossible to right after searching out characteristic point
Characteristic point carries out calculating process.By herein matrixing being carried out to characteristic point, characteristic point can be further processed.
To sum up, the present invention by smart mobile phone acceleration transducer is carried out with the separation of data, feature extraction, compile by feature
Code, a series of process such as characteristic point vectorization, effectively solve more complicated to smart mobile phone movement locus feature extraction
Problem, and coded quantization is carried out for wave character, key message is effectively stored, has greatly been saved smart mobile phone
Storage resource, feature coding can reduce characteristic point is normalized, and has saved the computing resource of smart mobile phone mobile phone,
And without by by extras, locally real-time movement locus Feature extraction and recognition can carried out it is adaptable to provide
The embedded devices such as the smart mobile phone limited by source.
Claims (4)
1. a kind of smart mobile phone locally in real time movement locus Feature extraction and recognition method it is characterised in that：The method is divided into
Training and two stages of identification；In the training stage：User carries smart mobile phone and does different actions, gathers separated intelligent mobile phone three
Axle acceleration sensor data；Wave crest and wave trough characteristic point is extracted to user's attitude behavior；Carry out two to Wave crest and wave trough characteristic point to enter
Coded quantization processed；Characteristic point vectorization after coding；And Wave crest and wave trough quantity identical action is carried out matrixing, gather multiple samples
This is trained in host computer, sets up user action characteristic standard storehouse；In cognitive phase：Transplanting user action characteristic standard storehouse is extremely
Smart mobile phone, when user carries smart mobile phone and does different actions, gathers sensing data in mobile phone terminal, slow in mobile phone terminal developing
Rush area, extract motion trace data, set up multithreading and extract segmentation feature point, mate user action characteristic standard storehouse, intelligent handss
Machine locally carries out feature point extraction in real time, and then identifying user action.
2. a kind of smart mobile phone according to claim 1 locally real-time movement locus Feature extraction and recognition method, its
It is characterised by：Smart mobile phone acceleration transducer is carried out separating along three axles.
3. a kind of smart mobile phone according to claim 1 locally real-time movement locus Feature extraction and recognition method, its
It is characterised by：Form using one piece of data segmentation extracts Wave crest and wave trough characteristic point to user's attitude behavior.
4. a kind of smart mobile phone according to claim 3 locally real-time movement locus Feature extraction and recognition method, its
It is characterised by：The form of described utilization one piece of data segmentation is specially：Identify crest and the ripple of one piece of data using sliding window
Paddy, using the data of the one section of fixed number producing as definite value sliding window, user movement status data is all in definite value sliding window
In mouthful, sliding window is divided into the sliding window of equal portions, carries out the inspection of Wave crest and wave trough data in this little sliding window
Survey, in each little sliding window, can have crest and trough, the order according to occurring is ranked up, crest and trough is respective
Compare, rear several troughs of the maximum front several crests of value and value minimum, as the characteristic point of each axle.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610732089.XA CN106406516A (en) | 2016-08-26 | 2016-08-26 | Local real-time movement trajectory characteristic extraction and identification method for smartphone |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610732089.XA CN106406516A (en) | 2016-08-26 | 2016-08-26 | Local real-time movement trajectory characteristic extraction and identification method for smartphone |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106406516A true CN106406516A (en) | 2017-02-15 |
Family
ID=58004828
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610732089.XA Pending CN106406516A (en) | 2016-08-26 | 2016-08-26 | Local real-time movement trajectory characteristic extraction and identification method for smartphone |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106406516A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108737623A (en) * | 2018-05-31 | 2018-11-02 | 南京航空航天大学 | The method for identifying ID of position and carrying mode is carried based on smart mobile phone |
| CN110151187A (en) * | 2019-04-09 | 2019-08-23 | 缤刻普达（北京）科技有限责任公司 | Body-building action identification method, device, computer equipment and storage medium |
| CN113283493A (en) * | 2021-05-19 | 2021-08-20 | Oppo广东移动通信有限公司 | Sample acquisition method, device, terminal and storage medium |
| CN113780447A (en) * | 2021-09-16 | 2021-12-10 | 郑州云智信安安全技术有限公司 | Sensitive data discovery and identification method and system based on flow analysis |
| WO2023178594A1 (en) * | 2022-03-24 | 2023-09-28 | 广东高驰运动科技股份有限公司 | Action counting method and apparatus, device, and storage medium |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101788861A (en) * | 2009-01-22 | 2010-07-28 | 华硕电脑股份有限公司 | Method and system for identifying three-dimensional motion |
| CN102246125A (en) * | 2008-10-15 | 2011-11-16 | 因文森斯公司 | Mobile devices with motion gesture recognition |
| CN102772211A (en) * | 2012-08-08 | 2012-11-14 | 中山大学 | Human movement state detection system and detection method |
| CN103345627A (en) * | 2013-07-23 | 2013-10-09 | 清华大学 | Action recognition method and device |
| CN103517118A (en) * | 2012-12-28 | 2014-01-15 | Tcl集团股份有限公司 | Motion recognition method and system for remote controller |
| CN103886323A (en) * | 2013-09-24 | 2014-06-25 | 清华大学 | Behavior identification method based on mobile terminal and mobile terminal |
| CN103984416A (en) * | 2014-06-10 | 2014-08-13 | 北京邮电大学 | Gesture recognition method based on acceleration sensor |
| CN104754111A (en) * | 2013-12-31 | 2015-07-01 | 北京新媒传信科技有限公司 | Control method for mobile terminal application and control device |
| CN104750386A (en) * | 2015-03-20 | 2015-07-01 | 广东欧珀移动通信有限公司 | Gesture recognition method and device |
| CN105159441A (en) * | 2015-07-28 | 2015-12-16 | 东华大学 | Autonomous motion identification technology based private coach smart band |
-
2016
- 2016-08-26 CN CN201610732089.XA patent/CN106406516A/en active Pending
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102246125A (en) * | 2008-10-15 | 2011-11-16 | 因文森斯公司 | Mobile devices with motion gesture recognition |
| CN101788861A (en) * | 2009-01-22 | 2010-07-28 | 华硕电脑股份有限公司 | Method and system for identifying three-dimensional motion |
| CN102772211A (en) * | 2012-08-08 | 2012-11-14 | 中山大学 | Human movement state detection system and detection method |
| CN103517118A (en) * | 2012-12-28 | 2014-01-15 | Tcl集团股份有限公司 | Motion recognition method and system for remote controller |
| CN103345627A (en) * | 2013-07-23 | 2013-10-09 | 清华大学 | Action recognition method and device |
| CN103886323A (en) * | 2013-09-24 | 2014-06-25 | 清华大学 | Behavior identification method based on mobile terminal and mobile terminal |
| CN104754111A (en) * | 2013-12-31 | 2015-07-01 | 北京新媒传信科技有限公司 | Control method for mobile terminal application and control device |
| CN103984416A (en) * | 2014-06-10 | 2014-08-13 | 北京邮电大学 | Gesture recognition method based on acceleration sensor |
| CN104750386A (en) * | 2015-03-20 | 2015-07-01 | 广东欧珀移动通信有限公司 | Gesture recognition method and device |
| CN105159441A (en) * | 2015-07-28 | 2015-12-16 | 东华大学 | Autonomous motion identification technology based private coach smart band |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108737623A (en) * | 2018-05-31 | 2018-11-02 | 南京航空航天大学 | The method for identifying ID of position and carrying mode is carried based on smart mobile phone |
| CN110151187A (en) * | 2019-04-09 | 2019-08-23 | 缤刻普达（北京）科技有限责任公司 | Body-building action identification method, device, computer equipment and storage medium |
| CN110151187B (en) * | 2019-04-09 | 2022-07-05 | 缤刻普达（北京）科技有限责任公司 | Body-building action recognition method and device, computer equipment and storage medium |
| CN113283493A (en) * | 2021-05-19 | 2021-08-20 | Oppo广东移动通信有限公司 | Sample acquisition method, device, terminal and storage medium |
| CN113780447A (en) * | 2021-09-16 | 2021-12-10 | 郑州云智信安安全技术有限公司 | Sensitive data discovery and identification method and system based on flow analysis |
| CN113780447B (en) * | 2021-09-16 | 2023-07-11 | 郑州云智信安安全技术有限公司 | Sensitive data discovery and identification method and system based on flow analysis |
| WO2023178594A1 (en) * | 2022-03-24 | 2023-09-28 | 广东高驰运动科技股份有限公司 | Action counting method and apparatus, device, and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106406516A (en) | Local real-time movement trajectory characteristic extraction and identification method for smartphone | |
| Yang et al. | Deep convolutional neural networks on multichannel time series for human activity recognition. | |
| Subetha et al. | A survey on human activity recognition from videos | |
| CN105956560B (en) | A kind of model recognizing method based on the multiple dimensioned depth convolution feature of pondization | |
| CN108287989B (en) | Sliding verification code man-machine identification method based on track | |
| CN109858406B (en) | Key frame extraction method based on joint point information | |
| CN108509859A (en) | A kind of non-overlapping region pedestrian tracting method based on deep neural network | |
| Xia et al. | An evaluation of deep learning in loop closure detection for visual SLAM | |
| CN104281853A (en) | Behavior identification method based on 3D convolution neural network | |
| CN110232308B (en) | Robot-following gesture track recognition method based on hand speed and track distribution | |
| CN105095829A (en) | Face recognition method and system | |
| Zhang et al. | IF-ConvTransformer: A framework for human activity recognition using IMU fusion and ConvTransformer | |
| CN111723662B (en) | Human body posture recognition method based on convolutional neural network | |
| CN106815578A (en) | A kind of gesture identification method based on Depth Motion figure Scale invariant features transform | |
| CN110674875A (en) | Pedestrian motion mode identification method based on deep hybrid model | |
| CN103440667A (en) | Automatic device for stably tracing moving targets under shielding states | |
| Liu et al. | Hand Gesture Recognition Based on Single‐Shot Multibox Detector Deep Learning | |
| CN106023249A (en) | Moving object detection method based on local binary similarity pattern | |
| CN110533699A (en) | The dynamic multiframe speed-measuring method of pixel variation based on optical flow method | |
| CN105160285A (en) | Method and system for recognizing human body tumble automatically based on stereoscopic vision | |
| Zhang et al. | Trajectory series analysis based event rule induction for visual surveillance | |
| CN113780140A (en) | Gesture image segmentation and recognition method and device based on deep learning | |
| CN105631462A (en) | Behavior identification method through combination of confidence and contribution degree on the basis of space-time context | |
| CN109029432B (en) | Human body action detection and identification method based on six-axis inertial sensing signal | |
| Wang | Data feature extraction method of wearable sensor based on convolutional neural network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20170215 |
|
| RJ01 | Rejection of invention patent application after publication |