CN107766930A - Based on the fuzzy equivalent ROM distance calculating methods for dividing group's SOM neurons of DTW clusters - Google Patents
Based on the fuzzy equivalent ROM distance calculating methods for dividing group's SOM neurons of DTW clusters Download PDFInfo
- Publication number
- CN107766930A CN107766930A CN201710795124.7A CN201710795124A CN107766930A CN 107766930 A CN107766930 A CN 107766930A CN 201710795124 A CN201710795124 A CN 201710795124A CN 107766930 A CN107766930 A CN 107766930A
- Authority
- CN
- China
- Prior art keywords
- group
- rom
- mrow
- dtw
- equivalent
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/043—Architecture, e.g. interconnection topology based on fuzzy logic, fuzzy membership or fuzzy inference, e.g. adaptive neuro-fuzzy inference systems [ANFIS]
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
- G01C21/00—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00
- G01C21/10—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 by using measurements of speed or acceleration
- G01C21/12—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 by using measurements of speed or acceleration executed aboard the object being navigated; Dead reckoning
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
- G01C21/00—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00
- G01C21/20—Instruments for performing navigational calculations
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/088—Non-supervised learning, e.g. competitive learning
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Remote Sensing (AREA)
- Radar, Positioning & Navigation (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Software Systems (AREA)
- Automation & Control Theory (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Mathematical Physics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Optimization (AREA)
- Pure & Applied Mathematics (AREA)
- Mathematical Analysis (AREA)
- Fuzzy Systems (AREA)
- Computational Mathematics (AREA)
- User Interface Of Digital Computer (AREA)
- Feedback Control In General (AREA)
Abstract
Obscured the invention discloses one kind based on DTW clusters and divide group's SOM neuron ROM equivalent distances computational methods, foundation is directed to data of the characteristic vector after the secondary Gauss covariance white noise of initial state Kalman filtering algorithm, obtains the sample sequence for dynamic time warping algorithm；Sample sequence is converted into acceleration signal acceleration amplitude SVM X and the amplitude of Y-axis, divides group's nerve Meta algorithm that SVM amplitudes are divided into different groups using fuzzy；The clustering with reference to corresponding to the group's value and relative group value of above-mentioned SVM (x) and the axial directions of SVM (y) two falls central point and focus point；Fall center of gravity with reference to above-mentioned clustering and center obtains ROM equivalent numerical values and calculates distance.The present invention can find out the spatial offset distance of different group's sensors, reduce end-to-end figure's sensor path time.
Description
Technical field
The present invention relates to sensor information processing and technical field of hand gesture recognition, more particularly to one kind to be returned based on dynamic time
The whole fuzzy ROM equivalent distances computational methods for dividing group's SOM neurons.
Background technology
Intelligent bracelet, there is wearable, small volume, multisensor, be wirelessly transferred, node can be abstracted into and be used for
Human perception network, you can with by both hands ring or multiple wrist strap sensor groups into limited wireless sensor network.Pass through nothing
Line transport module transmitting data in real time, wireless sensor network combination neuron topological structure, can be with dynamic monitoring network node
Mobility, regular movements handles human body multidimensional posture perception information.By the acceleration transducer of wrist strap sensor,
Angular-rate sensor carries out more figure's modelings, and the foundation of operational model, record are cooperateed with to the action trail of user's wrist strap sensor
Human body attitude behavior, can be to being tested and scenario analysis on human body attitude.
In the prior art, be used as the change that gesture recognition majority is direction and acceleration based on angle judge according to
According to.But angle variable quantityization processing can not carried out quantifying criterion and tracking to gesture recognition.
The defects of in order to overcome prior art, the present invention propose a kind of fuzzy based on dynamic time consolidation and divide group SOM
The ROM equivalent distances computational methods of neuron.Sensor node network is passed through SOM neutral nets member dynamic and examined by the present invention
Sensor node is surveyed, network topology is among dynamic change.Gestures direction is identified according to dynamic time warping, according to ROM etc.
Effect distance identification gesture deviation angle.Modeled by sensing data and combined with multi-channel intelligent controlling transmission, realize that human body moves
State perceives optimal regular movements distance detection.Improve detection human body attitude and perceive efficiency.
The content of the invention
The present invention proposes a kind of equivalent ROM distance calculating methods for being obscured based on DTW clusters and dividing group's SOM neurons, should
Method comprises the following steps：
A. sensor network, the synchronous motor message for obtaining human motion, institute are set up using at least two motion sensors
Stating motor message includes three positioner acceleration signals and three-dimensional angular velocity signal；
B. the characteristic vector of the motor message is established, the characteristic vector removes two by initial state Kalman filtering unit
After secondary Gauss covariance white noise, the sample sequence for dynamic time warping algorithm is obtained；
C. the sample sequence is converted into the first acceleration amplitude and the second acceleration amplitude, divides group neural using fuzzy
First computing unit first acceleration amplitude and second acceleration amplitude are assembled to be formed corresponding to the first group and the
Two groups；
D. it is worth with reference to the group of first group and second group, in the first group for obtaining first group
Heart point and first group's focus point, and the second group's central point and second group's focus point of second group；
E. according to first group central point, first group focus point, second group central point and described
ROM equivalent distances values are calculated in second group's focus point, and the ROM equivalent distances value is used to instruct human body dynamic sensing optimal
Regular movements distance detection selects.
Described obscured based on DTW clusters proposed by the present invention is divided in the equivalent ROM distance calculating methods of group's SOM neurons,
ROM equivalent distances value is calculated afterwards to further comprise：
F. weight is obtained according to SOM neurons criterion；The big ROM equivalent distances value of weight is preferential output valve；
G. node-locus length is tried to achieve according to weighted value；
H. output node path length and ROM equivalent distances values.
Described obscured based on DTW clusters proposed by the present invention is divided in the equivalent ROM distance calculating methods of group's SOM neurons,
The weighted value such as below equation represents：
Wherein,
In formula, H (i) represents current group's deviation amplitude,Represent equivalent distances average；ROM (i, k) is represented
ROM equivalent distances values；T represents current time；I represents current group；K represents current group's node.
Described obscured based on DTW clusters proposed by the present invention is divided in the equivalent ROM distance calculating methods of group's SOM neurons,
The node-locus length such as below equation represents：
S (t, i)=L (t, i) × ROM (t, (i, k)) × F (t, (i, k)) × NET (k)；
In formula, L (t, i) expression group i central point Fcm (i) and origin distance, NET (k) expression weights, F (t, (i,
K)) it is criterion information.
Described obscured based on DTW clusters proposed by the present invention is divided in the equivalent ROM distance calculating methods of group's SOM neurons,
Sensor angles variation characteristic value is extracted by screening when equivalent ROM distances calculate, quantitative analysis, angle value conversion distance,
Original orientation angles and mobile criterion are subjected to quantification treatment.
Described obscured based on DTW clusters proposed by the present invention is divided in the equivalent ROM distance calculating methods of group's SOM neurons,
Required gesture recognition sample can be screened by DTW after initial state Kalman filtering list, the sample sequence after screening
Group's SOM algorithms are divided to contribute to the packet by the posture of refinement again by fuzzy.
The present invention obtains motor message in wrist strap sensor and passes through primary KALMAN and secondary Gauss covariance white noise square
Handled after battle array filtering, on the basis of obtaining the azimuth motion eigenvector information of human body wrist, eigenvector information is moved
The consolidation of state time maps to obtain gesture orientation attitude information D (i), then for wireless network node based on wrist strap sensor etc.
Effect is fuzzy to divide group to obtain node R OM (i) values, is then combined with SOM nerves Meta algorithm and obtains wrist attitude information D (i) and ROM
(i) criterion obtains human hands motion equivalent distances information.The present invention can be carried out by DTW to required gesture recognition sample
Screening, the sample sequence after screening divide group's SOM algorithms to contribute to the packet by the posture of refinement again by fuzzy, it is possible to increase
Cluster analysis efficiency, angular quantification processing can be accelerated to polytomy variable dimensionality reduction after screening.It is different from conventional body's Attitude Tracking
Video flowing is needed to analyze attitude information, the present invention does not need criterion after video flowing acquisition process, reduces hardware cost.
In the present invention, DTW refers in dynamic time warping, and full name is Dynamic Time Warping；ROM refers to angle
Tracking, full name is Range of Motion；SOM is a kind of self-organizing (competitive type) neutral net, full name Self-
Organizing Maps。
Brief description of the drawings：
Fig. 1 is that fuzzy based on dynamic time consolidation divides group's SOM neuron ROM equivalent distances algorithms to be integrally in the present invention
System Organization Chart.
Fig. 2 is binodal point sensor ROM calculating figures in the present invention.
Fig. 3 is more piece point sensor ROM calculating figures in the present invention.
Fig. 4 is SOM neurons ordering chart in the present invention.
Embodiment
Wireless network node data acquisition fusion accelerometer and gyro sensor information in the present invention.Transported according to collection
Dynamic signal, is obtained in the characteristic vector constructed on each bracelet, and the data that MEMs modules obtain are sent to system, and the present invention exists
Spin matrix (rolling, pitching and course) is calculated on the basis of Eulerian angles, can be converted to and come from by this spin matrix
The acceleration information of people.
Wrist strap sensor will carry out first time processing to the data message after acceleration transducer data processing, and data are entered
Row quantification treatment, this is that software is realized, has feasibility.According to the motor message collected, filtered by polymorphic Kalman
Ripple, resulting information is handled, the motion of human upper limb wrist is detected according to the motor message of initial acquisition, and then obtained
Obtain the azimuth motion information of the hand of human body.Kalman filtering is filtering and the Predicting Technique of a kind of linear Gaussian Systems.It by
One processing model and a measurement model composition.Data processing principle is exactly mainly the fusion of double-channel data, has continued one
The technology of intelligent remote control systems and its method of the kind based on the fusion of both hands ring sensor, detection and estimating system state will biography
Sensor data are merged.In order to continuously estimate that hand motion state-space model is as follows：
X (k-1)=Mz (k-1)+u (k-1) (1)
Carry out the spatiality distribution processing carried out during another step data processing.X (k-1) represents to carry out at the first step data
Output result during reason, M are the quantization mapping tables of definition, and z (k-1) is the original output data of sensor, and u (k-1) is system
Noise.The present invention has continued a kind of intelligent remote control systems based on the fusion of both hands ring sensor and its method is melted
Close mapping table, y (k-1)=(x1(k-1),x2(k-1) twin-channel information aggregate), is represented.After y (k-1) is by merging mapping table,
Obtain gesture motion direction.The control in gesture motion direction, (B represents back) after BDFLR is represented respectively, under (D represents down), preceding
(F represents forward), left (L represents left) and the right side (R represents right).Setting only has two passages to send the following control word of state
Control character is just effective during symbol, otherwise invalid.Such as the control of only two bracelets, fusion posture processing judge character mode change because
Sub- Ri(i=12,3,4) it is
As shown in figure 1, equivalent obscure of the ROM based on dynamic time consolidation divides group's SOM nerve Meta algorithms：Wrist strap sensor three
Axle accelerometer samples to human body acceleration, and using sampling obtain acceleration signal be converted into acceleration signal amplitude as
The sample sequence of dynamic time warping.
D (i, j)=d (i, j)+min (D (i-1, j), D (i, j-1), D (i-1, j)) (3.1)
Wherein：
D (i, j) is the middle minimum cumulative distance of continuous acceleration signal amplitude in above-mentioned formula, and d (i, j) is continuously to accelerate
Spend several inner distances in Europe of signal amplitude.
Incidence coefficient：
Wherein, ε ∈ (0,2) be adjustable parameter according to sensor 1, sensor 2, sensor 3,4 different values of sensor not
Together.
Sensor 1, sensor 2, sensor 3, the acceleration signal acceleration amplitude SVM of sensor 4, lose serious Z axis
Wave the amplitude trend for not influenceing X and Y-axis, therefore two values of SVM (x) and SVM (y) are only recorded, to acceleration signal amplitude
(SVM) end-point detection is carried out, predicts wrist motion state.
As shown in Fig. 2 divide group's nerve Meta algorithm according to DTW incidence coefficients by sensor 1, the number of sensor 2 using fuzzy
According to 3 groups are divided into, its masses' central point (Fcm1, Fcm23) is obtained respectively.Each group of sensor in situation 1 is obtained in clustering
Heart point, and obtain its position of centre of gravity (GM1, GM2).
As shown in figure 3, divide group's nerve Meta algorithm according to DTW incidence coefficients by sensor 1, sensor 2, sensing using fuzzy
Device 3, the data of sensor 4 are divided into six groups, obtain its masses' central point (Fcm1, Fcm2, Fcm3, Fcm4) respectively.To situation 2
In each group of sensor obtain clustering central point, and obtain its position of centre of gravity (GM1, GM2, GM3, GM4).
As shown in Fig. 2 the anti-triangle of SVM (x) and SVM (y) value of position of centre of gravity is calculated respectively：
Both subtract each other to obtain equivalent ROM：
ROM (1,1)=θG(1+1)-θG(1) (6)
As shown in figure 3, the SVM (x) and SVM (y) values corresponding to position of centre of gravity (GM1, GM2, GM3, GM4) are calculated respectively
Anti- triangle：
Both subtract each other to obtain equivalent ROM (i)：
ROM (i, k)=θG(k+i)-θG(k) (9)
By the equivalent ROM (t, (i, k)) and ROM of the continuous sequence continuously obtained (t+ Δs t, (i, k)) difference, and threshold value
Change obtains:
U (t, (i, k))=ROM (t+ Δs t, (i, k))-ROM (t, (i, k)) (10)
It is differential threshold that ξ, which is represented,.F (t, (i, k)) is criterion information.
SOM is a kind of self-organizing (competitive type) neutral net, the structure of self-organizing (competitive type) neutral net and its study
The characteristics of rule has oneself compared with other neutral nets.In network structure, it is usually to be made up of input layer and competition layer
Two-tier network；Each neuron is realized and is bi-directionally connected between two layers.
As shown in figure 4, it is made up of input layer, competition layer and output layer.Input layer be position of centre of gravity (GM1, GM2, GM3,
GM4) different wrist strap sensor collection post processing states.Vied each other between network neural member in the hope of being activated, as a result each
Moment only has an output neuron to be activated.This neuron being activated referred to as competition triumph neuron, and other nerves
The state of member is suppressed.Weights are big or what fluctuation was big will preferentially be exported NET (t, k).The first output as figure's detection
Value, successively the second output valve, the 3rd output valve, and then show figure's change.
Relative distance S (t, i) represents wrist strap sensor node path length:
S (t, i)=L (t, i) × ROM (t, (i, k)) × F (t, (i, k)) × NET (k) (14)
L (t, i) is represented such as the distance of Fcm (i) and origin in Fig. 3 or Fig. 4.
The equivalent fuzzy processes for dividing group's SOM nerves Meta algorithm to obtain criterion information of ROM based on dynamic time consolidation are specific
Comprise the following steps：
Step 1：The characteristic vector of each bracelet passes through the secondary Gauss covariance white noise of initial state Kalman filtering algorithm
After sound, the sample sequence for dynamic time warping algorithm is obtained；
Step 2：Human body acceleration is sampled by bracelet sensor triaxial accelerometer, and the acceleration that sampling is obtained
Degree signal is converted into acceleration signal acceleration amplitude SVM；
Step 3：Record sensor 1, sensor 2, sensor 3, the acceleration signal acceleration amplitude SVM of sensor 4, damage
The amplitude trend for not influenceing X and Y-axis of waving of serious Z axis is lost, therefore only records two values of SVM (x) and SVM (y)；
Step 4：Divide group's nerve Meta algorithm by the data of sensor 1, sensor 2, sensor 3, sensor 4 point using fuzzy
Into six groups, obtain respectively its masses central point Fcm (i) (i.e. Fcm1, Fcm2, Fcm3, Fcm4) and central point and origin away from
From L (t, i)；
Step 5：Clustering central point is obtained to each group of sensor in situation 1, and obtain its position of centre of gravity (GA1, GA2,
GA3,GA4).Clustering central point is obtained to each group of sensor in situation 2, and obtain its position of centre of gravity (GB1, GB2, GB3,
GB4)；
Step 6：Calculation process is carried out to its calculated position of centre of gravity, obtains ROM (i) equivalent numerical values；
Step 7：Weight NET (t, k) is obtained according to SOM neuron criterions；
Step 8：Try to achieve node-locus length S (t, i)；
Step 9：By resulting S (t, i) and ROM (t, (i, k)) equivalence value by Serial Port Transmission to mobile terminal and
Computer terminal.
The protection content of the present invention is not limited to above example.Under the spirit and scope without departing substantially from inventive concept, this
Art personnel it is conceivable that change and advantage be all included in the present invention, and using appended claims as protect
Protect scope.
Claims (6)
- It is 1. a kind of based on the fuzzy equivalent ROM distance calculating methods for dividing group's SOM neurons of DTW clusters, it is characterised in that this method Comprise the following steps：A. sensor network, the synchronous motor message for obtaining human motion, the fortune are set up using at least two motion sensors Dynamic signal includes three positioner acceleration signals and three-dimensional angular velocity signal；B. the characteristic vector of the motor message is established, the characteristic vector removes secondary height by initial state Kalman filtering unit After this covariance white noise, the sample sequence for dynamic time warping algorithm is obtained；C. the sample sequence is converted into the first acceleration amplitude and the second acceleration amplitude, divides group's neuron meter using fuzzy Unit is calculated to assemble first acceleration amplitude and second acceleration amplitude to form corresponding first group and second group Fall；D. it is worth with reference to the group of first group and second group, obtains first group's central point of first group With the second group's central point and second group's focus point of first group's focus point, and second group；E. according to first group central point, first group focus point, second group central point and described second ROM equivalent distances values are calculated in group's focus point, and the ROM equivalent distances value is used to instruct the optimal regular movements of human body dynamic sensing Distance detection selection.
- 2. as claimed in claim 1 based on the fuzzy equivalent ROM distance calculating methods for dividing group's SOM neurons of DTW clusters, it is special Sign is, ROM equivalent distances value is calculated and further comprises afterwards：F. weight is obtained according to SOM neurons criterion；The big ROM equivalent distances value of weight is preferential output valve；G. node-locus length is tried to achieve according to weighted value；H. output node path length and ROM equivalent distances values.
- 3. as claimed in claim 1 based on the fuzzy equivalent ROM distance calculating methods for dividing group's SOM neurons of DTW clusters, it is special Sign is that the weighted value such as below equation represents：<mrow> <mi>N</mi> <mi>E</mi> <mi>T</mi> <mrow> <mo>(</mo> <mi>t</mi> <mo>,</mo> <mi>k</mi> <mo>)</mo> </mrow> <mo>=</mo> <mi>M</mi> <mi>a</mi> <mi>x</mi> <mo>{</mo> <mi>H</mi> <mrow> <mo>(</mo> <mi>i</mi> <mo>)</mo> </mrow> <mo>&times;</mo> <msup> <mrow> <mo>&lsqb;</mo> <mover> <mrow> <mi>R</mi> <mi>O</mi> <mi>M</mi> </mrow> <mo>&OverBar;</mo> </mover> <mrow> <mo>(</mo> <mi>i</mi> <mo>,</mo> <mi>k</mi> <mo>)</mo> </mrow> <mo>-</mo> <mi>R</mi> <mi>O</mi> <mi>M</mi> <mrow> <mo>(</mo> <mi>i</mi> <mo>,</mo> <mi>k</mi> <mo>)</mo> </mrow> <mo>&rsqb;</mo> </mrow> <mn>2</mn> </msup> <mo>}</mo> <mo>;</mo> </mrow>Wherein,In formula, H (i) represents current group's deviation amplitude,Represent equivalent distances average；ROM (i, k) represents ROM etc. Imitate distance value；T represents current time；I represents current group；K represents current group's node.
- 4. as claimed in claim 3 based on the fuzzy equivalent ROM distance calculating methods for dividing group's SOM neurons of DTW clusters, it is special Sign is that the node-locus length such as below equation represents：S (t, i)=L (t, i) × ROM (t, (i, k)) × F (t, (i, k)) × NET (k)；In formula, L (t, i) represents group i central point Fcm (i) and origin distance, and NET (k) represents weight, and F (t, (i, k)) is Criterion information.
- 5. as claimed in claim 1 based on the fuzzy equivalent ROM distance calculating methods for dividing group's SOM neurons of DTW clusters, it is special Sign is, extracts sensor angles variation characteristic value, quantitative analysis by screening when equivalent ROM distances calculate, angle value turns Change distance, original orientation angles and mobile criterion are subjected to quantification treatment.
- 6. as claimed in claim 1 based on the fuzzy equivalent ROM distance calculating methods for dividing group's SOM neurons of DTW clusters, it is special Sign is, required gesture recognition sample can be screened by DTW after initial state Kalman filtering list, the sample after screening This sequence divides group's SOM algorithms to contribute to the packet by the posture of refinement again by fuzzy.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710795124.7A CN107766930B (en) | 2017-09-06 | 2017-09-06 | Equivalent ROM distance calculation method based on DTW cluster fuzzy clustering SOM neurons |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710795124.7A CN107766930B (en) | 2017-09-06 | 2017-09-06 | Equivalent ROM distance calculation method based on DTW cluster fuzzy clustering SOM neurons |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107766930A true CN107766930A (en) | 2018-03-06 |
| CN107766930B CN107766930B (en) | 2021-03-26 |
Family
ID=61265414
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710795124.7A Active CN107766930B (en) | 2017-09-06 | 2017-09-06 | Equivalent ROM distance calculation method based on DTW cluster fuzzy clustering SOM neurons |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107766930B (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108712760A (en) * | 2018-03-29 | 2018-10-26 | 北京邮电大学 | High-throughput relay selection method based on random Learning Automata and fuzzy algorithmic approach |
| CN110095120A (en) * | 2019-04-03 | 2019-08-06 | 河海大学 | Biology of the Autonomous Underwater aircraft under ocean circulation inspires Self-organizing Maps paths planning method |
| CN110319829A (en) * | 2019-07-08 | 2019-10-11 | 河北科技大学 | Unmanned aerial vehicle flight path planing method based on adaptive polymorphic fusion ant colony algorithm |
| CN110415336A (en) * | 2019-07-12 | 2019-11-05 | 清华大学 | High-precision human posture method for reconstructing and system |
| CN113065604A (en) * | 2021-04-15 | 2021-07-02 | 北京理工大学 | Air target grouping method based on DTW-DBSCAN algorithm |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7855684B2 (en) * | 2007-11-15 | 2010-12-21 | Samsung Electronics Co., Ltd. | Method and system for locating sensor node in sensor network using distance determining algorithm |
| WO2013096954A1 (en) * | 2011-12-23 | 2013-06-27 | The Trustees Of Dartmouth College | Wearable computing device for secure control of physiological sensors and medical devices, with secure storage of medical records, and bioimpedance biometric |
| CN103543826A (en) * | 2013-07-30 | 2014-01-29 | 广东工业大学 | Method for recognizing gesture based on acceleration sensor |
| CN106500695A (en) * | 2017-01-05 | 2017-03-15 | 大连理工大学 | A kind of human posture recognition method based on adaptive extended kalman filtering |
-
2017
- 2017-09-06 CN CN201710795124.7A patent/CN107766930B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7855684B2 (en) * | 2007-11-15 | 2010-12-21 | Samsung Electronics Co., Ltd. | Method and system for locating sensor node in sensor network using distance determining algorithm |
| WO2013096954A1 (en) * | 2011-12-23 | 2013-06-27 | The Trustees Of Dartmouth College | Wearable computing device for secure control of physiological sensors and medical devices, with secure storage of medical records, and bioimpedance biometric |
| CN103543826A (en) * | 2013-07-30 | 2014-01-29 | 广东工业大学 | Method for recognizing gesture based on acceleration sensor |
| CN106500695A (en) * | 2017-01-05 | 2017-03-15 | 大连理工大学 | A kind of human posture recognition method based on adaptive extended kalman filtering |
Non-Patent Citations (3)
| Title |
|---|
| CHIH-YI CHIU ET AL.: "Content-Based Retrieval for Human Motion Data", 《JOURNAL OF VISUAL COMMUNICATION AND IMAGE REPRESENTATION》 * |
| HAIPENG QU ET AL.: "Localization with Single Stationary Anchor for Mobile Node in Wireless Sensor Networks", 《HINDAWI》 * |
| 张浩 等: "加权DTW距离的自动步态识别", 《中国图象图形学报》 * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108712760A (en) * | 2018-03-29 | 2018-10-26 | 北京邮电大学 | High-throughput relay selection method based on random Learning Automata and fuzzy algorithmic approach |
| CN108712760B (en) * | 2018-03-29 | 2019-11-19 | 北京邮电大学 | High-throughput relay selection method based on random Learning Automata and fuzzy algorithmic approach |
| CN110095120A (en) * | 2019-04-03 | 2019-08-06 | 河海大学 | Biology of the Autonomous Underwater aircraft under ocean circulation inspires Self-organizing Maps paths planning method |
| CN110319829A (en) * | 2019-07-08 | 2019-10-11 | 河北科技大学 | Unmanned aerial vehicle flight path planing method based on adaptive polymorphic fusion ant colony algorithm |
| CN110319829B (en) * | 2019-07-08 | 2022-11-18 | 河北科技大学 | Unmanned aerial vehicle flight path planning method based on self-adaptive polymorphic fusion ant colony algorithm |
| CN110415336A (en) * | 2019-07-12 | 2019-11-05 | 清华大学 | High-precision human posture method for reconstructing and system |
| CN113065604A (en) * | 2021-04-15 | 2021-07-02 | 北京理工大学 | Air target grouping method based on DTW-DBSCAN algorithm |
| CN113065604B (en) * | 2021-04-15 | 2022-10-21 | 北京理工大学 | Air target grouping method based on DTW-DBSCAN algorithm |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107766930B (en) | 2021-03-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107766930A (en) | Based on the fuzzy equivalent ROM distance calculating methods for dividing group's SOM neurons of DTW clusters | |
| Liu et al. | Fusion of magnetic and visual sensors for indoor localization: Infrastructure-free and more effective | |
| CN112153736B (en) | Personnel action identification and position estimation method based on channel state information | |
| Gu et al. | Paws: Passive human activity recognition based on wifi ambient signals | |
| CN108245880A (en) | Body-sensing detection method for visualizing and system based on more wearing annulus sensor fusions | |
| CN109579853A (en) | Inertial navigation indoor orientation method based on BP neural network | |
| Asadi et al. | A Particle Swarm Optimization-based washout filter for improving simulator motion fidelity | |
| EP2624171B1 (en) | Method and node for analysing periodic movements | |
| Yang et al. | Motion-pattern recognition system using a wavelet-neural network | |
| Hafeez et al. | Multi-fusion sensors for action recognition based on discriminative motion cues and random forest | |
| CN107454970A (en) | A kind of System and method for of movement locus collection and analysis based on ball game | |
| CN108827290A (en) | A kind of human motion state inverting device and method | |
| Hajjej et al. | Deep human motion detection and multi-features analysis for smart healthcare learning tools | |
| CN106970705A (en) | Motion capture method, device and electronic equipment | |
| Lisowska et al. | Cross-cohort Evaluation of Machine Learning Approaches to Fall Detection from Accelerometer Data. | |
| KR102108180B1 (en) | Method and wearable device for providing feedback on exercise | |
| Chen et al. | An Online Prediction and Trajectory Tracking Method for Human Activity Recognition | |
| Wang | Data feature extraction method of wearable sensor based on convolutional neural network | |
| Yu et al. | Deep learning-based vibration signal personnel positioning system | |
| Zhang et al. | A fall detection study based on neural network algorithm using AHRS | |
| Abougarair et al. | Analysis of Mobile Accelerometer Sensor Movement Using Machine Learning Algorithm | |
| Mekruksavanich et al. | Free-Weight Exercise Activity Recognition using Deep Residual Neural Network based on Sensor Data from In-Ear Wearable Devices | |
| Guo et al. | Human activity recognition based on two-dimensional acoustic arrays | |
| Zhang | [Retracted] Application of Intelligent Sensor Network in the Assessment of Table Tennis Teaching and Training Intensity, Training Volume, and Physical Fitness | |
| Agius et al. | A wearable wireless sensing system for capturing human arm motion |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CP02 | Change in the address of a patent holder |
Address after: 200241 No. 500, Dongchuan Road, Shanghai, Minhang District Patentee after: EAST CHINA NORMAL University Patentee after: SHANGHAI MINGFENG INFORMATION TECHNOLOGY Co.,Ltd. Address before: 200062 No. 3663, Putuo District, Shanghai, Zhongshan North Road Patentee before: EAST CHINA NORMAL University Patentee before: SHANGHAI MINGFENG INFORMATION TECHNOLOGY Co.,Ltd. |
|
| CP02 | Change in the address of a patent holder |