CN107451594A - A kind of various visual angles Approach for Gait Classification based on multiple regression - Google Patents
A kind of various visual angles Approach for Gait Classification based on multiple regression Download PDFInfo
- Publication number
- CN107451594A CN107451594A CN201710572423.4A CN201710572423A CN107451594A CN 107451594 A CN107451594 A CN 107451594A CN 201710572423 A CN201710572423 A CN 201710572423A CN 107451594 A CN107451594 A CN 107451594A
- Authority
- CN
- China
- Prior art keywords
- mrow
- gait
- msub
- visual angles
- multiple regression
- 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
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- 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
- G06N3/045—Combinations of networks
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/07—Target detection
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- Molecular Biology (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a kind of various visual angles Approach for Gait Classification based on multiple regression, comprise the following steps:Step 1: concentrating extraction human body contour outline from gait data to be sorted, the interior point of construction represents and two kinds of human region images of boundary representation, and further generation wheel hub energy diagram;Step 2: the wheel hub energy diagram based on previous step, calculates multicycle mixing gait energy matrix corresponding to each gait image sequence as gait feature;Step 3: various visual angles gait classification is converted into a multiple regression problem, and constructs convolutional neural networks and solved.The various visual angles Approach for Gait Classification based on multiple regression of the present invention has used multicycle mixing gait energy matrix to split without accurate gait cycle as gait feature, reduce the dependence to gait cycle segmentation.Meanwhile various visual angles gait classification problem is converted into multiple regression problem by the present invention, is solved by convolutional neural networks, improves gait classification accuracy.
Description
Technical field
It is specifically a kind of based on the more of multiple regression the invention belongs to the biological characteristic processing technology field in pattern-recognition
Visual angle Approach for Gait Classification.
Background technology
Gait classification is one of biological characteristic treatment technology most potential under remote testing conditions, and it can be in dress ornament
Identification and lower limb health status monitoring are realized with the posture walked under conditions of the change of the factor such as visual angle according to people.Gait
The basic step of classification is:First, concentrated by moving object detection from gait data to be sorted and separate gait image.
Secondly, the detection of gait cycle is carried out, is partitioned into a series of periodicity gait image sequence, and therefrom extract gait feature.
The selection of gait feature and accurate extraction are one of most important links in gait classification, and the classification for directly affecting follow-up is correct
Rate.Finally, design appropriate grader and carry out gait classification.Existing gait classification algorithm is when extracting gait feature, it is necessary to elder generation
Gait cycle segmentation is carried out, there is serious dependence to factors such as time and leg speeds, limit the application of gait classification and push away
Extensively.
Compared to the treatment technology of the biological characteristics such as face, fingerprint and iris, because the development time of gait processing is relative
It is shorter, in the stage of current gait classification with identification technology still in research, lack the method and theory of system.
The content of the invention
In order to solve above mentioned problem present in existing gait classification technology, the invention provides one kind to be based on multiple regression
Various visual angles Approach for Gait Classification, including following three step:
Step 1. concentrates extraction human body contour outline from gait data to be sorted, and the interior point of construction represents and two kinds of people of boundary representation
Body region image, and further generate wheel hub energy diagram;
First, by moving object detection, the human region that interior point represents is obtained:
Whether I (X, Y) value denotation coordination belongs to the interior point of human region for the pixel of (X, Y), i.e., is in the picture
It is no fall in human region;
Secondly, the human region represented according to interior point, the barycentric coodinates in current human region are calculated:
N hereinPRepresent the number of pixels of human region in image:
Again, edge extracting is further carried out, obtains the human region of boundary representation:
Whether E (X, Y) value denotation coordination belongs to the boundary point of human region for the pixel of (X, Y), i.e., in the picture
Whether fall on edges of regions line;
Finally, according to the human region (formula 4) and barycentric coodinates (formula 2) of boundary representation, connect each boundary point with
Center of gravity, the wheel hub energy diagram for characterizing the features such as the off-center point of human body contour outline is established, it is as follows:
Whether whether L (X, Y) value denotation coordination is located on wheel hub for the pixel of (X, Y), i.e., fall in body contour line
On point and center of gravity line on.
Wheel hub energy diagram of the step 2. based on previous step, calculates multicycle mixing step corresponding to each gait image sequence
State energy matrix is as gait feature;
Given one includes NFThe gait image sequence of two field picture, the width of each two field picture is W, is highly H, then corresponds to
Multicycle mixing gait energy matrix MH,WIt is defined as follows:
Here m (i, j) is the multicycle to mix gait energy matrix MH,WThe element of i-th row jth row, i ∈ [1, H], j ∈
[1,W].The human body contour outline information at each moment during multicycle mixing gait energy matrix reflects people on foot, and with
Time change, human body contour outline relative to center of gravity changing rule.
Various visual angles gait classification is converted into a multiple regression problem by step 3., and is constructed convolutional neural networks and asked
Solution;
In the gait data acquisition of reality, no matter using what shooting visual angle, part gait information is always caused to be hidden
Gear, can not extract complete gait feature.The present invention is on the basis of 90 degree of visual angles (i.e. people's direction of travel is parallel with the plane of delineation)
Data, gait feature vector is extracted, and construct convolutional neural networks and realize gait classification.Idiographic flow is as follows:
First, it is more corresponding to calculating according to the gait image sequence at 90 degree of visual angles of each object in data set to be identified
Cycle mixing gait energy matrix MH,W, and construct gait feature vector:
Here i ∈ [1, W],It is MH,WIt is vectorial corresponding to each row.
Secondly, a convolutional neural networks model φ is defined, realizes and step is extracted from the gait image sequence of visual angle
State characteristic vector.Model φ first layer is input layer, the human region I represented for inputting interior point;Last layer is output
Layer, exports corresponding gait feature vector prediction result;Centre is combined for a series of convolutional layer and pond layer, and one complete
Articulamentum.
y*=φ (I, θ) (8)
Finally, loss function is defined, and model φ is trained to obtain optimal model parameters.The optimization of parameter θ exists
Chinese Academy of Sciences's automation research CASIADatasetB data sets on carried out realization and various visual angles gait classification experiment.
Here, | | | | represent Euclidean distance.
Brief description of the drawings
Fig. 1 is the human region schematic diagram that interior point represents;
Fig. 2 is the human region schematic diagram that boundary point represents;
Fig. 3 is the various visual angles Gait Recognition flow chart based on multiple regression;
Fig. 4 is convolutional neural networks schematic diagram used;
Fig. 5 is the accumulative matching characteristic curve for testing three kinds of methods in 1;Wherein, GrassMannian methods have merged more
Viewing matrix represents and a kind of stochastic kernel extreme learning machine;VTM methods make use of a kind of standards of grading framework.
Fig. 6 is the accumulative matching characteristic curve for testing three kinds of methods in 2.Wherein, GrassMannian methods have merged more
Viewing matrix represents and a kind of stochastic kernel extreme learning machine;VTM methods make use of a kind of standards of grading framework.
Embodiment
The invention will be further described below in conjunction with the accompanying drawings.
Relative to prior art, the present invention has following features:
First, after extraction human body contour outline is concentrated from gait data to be sorted, basic gait pattern is described using wheel hub energy diagram
Information.
As depicted in figs. 1 and 2, the present invention obtains interior point and represented respectively first by moving object detection and edge extracting
With the human region schematic diagram of boundary representation.Then the human region represented according to interior point calculates focus point, and further connects
All boundary points and focus point generate wheel hub energy diagram.For the region of any convex polygon, wheel hub energy diagram represents with interior point
Region be completely superposed, but for human region non-convex under normal circumstances, the region that profile energy diagram and interior point represent is deposited
In significant difference, in the case of preferably characterizing different gaits, the change details in human body contour outline region.
2nd, the multicycle is selected to mix gait energy matrix as gait feature.
The various visual angles Approach for Gait Classification based on multiple regression of the present invention has used the multicycle to mix gait energy matrix
As gait feature, split without accurate gait cycle, reduce the dependence to gait cycle segmentation.Mixed using the multicycle
Gait energy matrix is closed as gait feature, there is some following advantage:
(1) split without accurate gait cycle, reduce what gait classification precision and accuracy were split to gait cycle
Dependence;
(2) multicycle mixing gait energy matrix is accumulation of the body gait characteristic in time-domain, can reflect that people walks
The gait inherent difference such as road custom and frequency;
(3) multicycle mixing gait energy matrix is also real embodiment of the body gait characteristic in spatial domain, the matrix
Each row can reflect the gait details in certain space field.
3rd, various visual angles gait classification is converted into multiple regression problem, and constructs convolutional neural networks and solved.
On the basis of the multicycle mixes gait energy matrix, present applicant proposes a kind of new various visual angles gait classification side
Method.Its basic thought is:Regard gait classification as a multiple regression problem, using the gait image sequence at 90 degree of visual angles as base
Standard, calculate multicycle mixing gait energy matrix MH,W;Then split-matrix MH,WTo generate the reference data of gait feature vector;
Last training convolutional neural networks carry out gait classification.
As shown in figure 3, the comprehensive utilization correlation theory such as multiple regression and convolutional neural networks, the various visual angles that the application proposes
Approach for Gait Classification flow is as follows:
(1) average multicycle mixing gait energy matrix is calculated.That asks for input corresponds to 90 degree of visual angle gaits
A series of mixing gait energy matrix M of multicycles of image sequenceH,WAverage value:
(2) reference data of gait feature vector is generated.Pass through decompositionUsing its it is each row as an independence to
Amount, to construct the reference data of gait feature vector.
(3) convolutional neural networks are constructed.As shown in figure 4, the neutral net that the present invention uses is divided into 7 layers, input therein
The human body contour outline area image that point represents in layer processing, size is 32*32 pixels;Output layer output is corresponding to current gait figure
As the gait feature vector of sequence;Middle hidden layer includes two convolutional layers and two pond layers, and one is used to classify
Full articulamentum.
(4) loss function is defined, and the study and training of convolutional neural networks are carried out using back-propagation algorithm.The step
Basic ideas be:The human region image training data that interior point represents is input in convolutional neural networks, by each hidden
Layer is hidden, and eventually arrives at output layer and obtains gait feature vector, completes propagated forward process;The loss defined according to formula (9)
Function calculates the error between epicycle output valve and actual value, and by the error from output layer to hidden layer backpropagation, until
Travel to input layer;During backpropagation, the parameter value for the adjustment hidden layer that iterated according to error, until convergence.
(5) reality according to obtained by everyone the gait feature vector reference data and (4) step of gained in (2) step
Gait feature vector, carry out gait classification.The nearest neighbor method that the present invention employs classics in experimental design is classified.
Experimental design:
Experiment 1:The gait data of all 90 degree of visual angle normal conditions is concentrated to be divided using the B in CASIA gait datas storehouse
Class accuracy compares
Training data acquisition methods:Randomly select under normal condition everyone 90 degree of visual angle gait datas 50% is used to instruct
Practice, remaining data are used to test.As shown in figure 4, transverse axis is rank values, the longitudinal axis is the accumulative matching characteristic curve of experimental result
The accuracy of gait classification.
Experiment 2:Whole normal condition data are concentrated to carry out classification accuracy rate comparison using the B in CASIA gait datas storehouse
Training data acquisition methods:Randomly select in whole normal condition data (including data under all 11 visual angles)
50% data at everyone 90 degree of visual angles are used to train, and remaining 50% visual angle is 90 degree of data, and other perspective datas
For testing.The accumulative matching characteristic curve of experimental result as shown in figure 5, transverse axis is rank values, the longitudinal axis be gait classification just
True rate.
Claims (4)
1. a kind of various visual angles Approach for Gait Classification based on multiple regression, including following three step:
Step 1. concentrates extraction human body contour outline from gait data to be sorted, and the interior point of construction represents and the Liang Zhong human bodies area of boundary representation
Area image, and further generate wheel hub energy diagram;
Step 2. calculates multicycle mixing gait energy corresponding to each gait image sequence based on described wheel hub energy diagram
Matrix is as gait feature;
Various visual angles gait classification is converted into a multiple regression problem by step 3., and is constructed convolutional neural networks and solved.
2. the various visual angles Approach for Gait Classification based on multiple regression as claimed in claim 1, it is characterised in that:
Step 1 is specially:
First, by moving object detection, the human region that interior point represents is obtained:
Whether whether I (X, Y) value denotation coordination belongs to the interior point of human region for the pixel of (X, Y), i.e., fall in the picture
In human region;
Secondly, the human region represented according to interior point, the barycentric coodinates in current human region are calculated:
<mrow>
<mfenced open = "{" close = "">
<mtable>
<mtr>
<mtd>
<mrow>
<msub>
<mi>x</mi>
<mi>c</mi>
</msub>
<mo>=</mo>
<mfrac>
<mn>1</mn>
<msub>
<mi>N</mi>
<mi>P</mi>
</msub>
</mfrac>
<munder>
<mo>&Sigma;</mo>
<mrow>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
</mrow>
</munder>
<mi>x</mi>
<mo>&CenterDot;</mo>
<mi>I</mi>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<msub>
<mi>y</mi>
<mi>c</mi>
</msub>
<mo>=</mo>
<mfrac>
<mn>1</mn>
<msub>
<mi>N</mi>
<mi>P</mi>
</msub>
</mfrac>
<munder>
<mo>&Sigma;</mo>
<mrow>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
</mrow>
</munder>
<mi>y</mi>
<mo>&CenterDot;</mo>
<mi>I</mi>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
</mrow>
</mtd>
</mtr>
</mtable>
</mfenced>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>2</mn>
<mo>)</mo>
</mrow>
</mrow>
N hereinPRepresent the number of pixels of human region in image:
<mrow>
<msub>
<mi>N</mi>
<mi>P</mi>
</msub>
<mo>=</mo>
<munder>
<mo>&Sigma;</mo>
<mrow>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
</mrow>
</munder>
<mi>I</mi>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>3</mn>
<mo>)</mo>
</mrow>
</mrow>
Again, edge extracting is further carried out, obtains the human region of boundary representation:
Whether E (X, Y) value denotation coordination belongs to the boundary point of human region for the pixel of (X, Y), i.e., in the picture whether
Fall on edges of regions line;
Finally, according to the human region and barycentric coodinates of boundary representation, each boundary point and center of gravity are connected, establishes and characterizes people
The wheel hub energy diagram of the features such as the off-center point of body profile, it is as follows:
Whether whether L (X, Y) value denotation coordination is located on wheel hub for the pixel of (X, Y), i.e., fall on body contour line
On point and center of gravity line.
3. the various visual angles Approach for Gait Classification based on multiple regression as claimed in claim 1, it is characterised in that:
Step 2 is specially:
Given one includes NFThe gait image sequence of two field picture, the width of each two field picture is W, is highly H, then corresponding more
Cycle mixing gait energy matrix MH,WIt is defined as follows:
<mrow>
<mi>m</mi>
<mrow>
<mo>(</mo>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfrac>
<mn>1</mn>
<msub>
<mi>N</mi>
<mi>F</mi>
</msub>
</mfrac>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>k</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mrow>
<mi>k</mi>
<mo>=</mo>
<msub>
<mi>N</mi>
<mi>F</mi>
</msub>
</mrow>
</munderover>
<msub>
<mi>L</mi>
<mi>k</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
<mo>)</mo>
</mrow>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>6</mn>
<mo>)</mo>
</mrow>
</mrow>
Here m (i, j) is the multicycle to mix gait energy matrix MH,WThe element of i-th row jth row, i ∈ [1, H], j ∈ [1,
W]。
4. the various visual angles Approach for Gait Classification based on multiple regression as claimed in claim 1, it is characterised in that:
Step 3 is specially:
The data on the basis of 90 degree of visual angles, gait feature vector is extracted, and construct convolutional neural networks and realize gait classification, specifically
Flow is as follows:
First, according to the gait image sequence at 90 degree of visual angles of each object in data set to be identified, the multicycle corresponding to calculating
Mix gait energy matrix MH,W, and construct gait feature vector:
<mrow>
<mi>y</mi>
<mo>=</mo>
<msup>
<mrow>
<mo>(</mo>
<mo>...</mo>
<mo>,</mo>
<msubsup>
<mi>y</mi>
<mi>i</mi>
<mi>T</mi>
</msubsup>
<mo>,</mo>
<mo>...</mo>
<mo>)</mo>
</mrow>
<mi>T</mi>
</msup>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>7</mn>
<mo>)</mo>
</mrow>
</mrow>
Here i ∈ [1, W],It is MH,WIt is vectorial corresponding to each row.
Secondly, a convolutional neural networks model φ is defined, realizes and gait spy is extracted from the gait image sequence of visual angle
Sign vector.Model φ first layer is input layer, the human region I represented for inputting interior point;Last layer is output layer, defeated
Go out corresponding gait feature vector prediction result;Centre is combined for a series of convolutional layer and pond layer, and a full connection
Layer.
y*=φ (I, θ) (8)
Finally, loss function is defined, and model φ is trained to obtain optimal model parameters,
<mrow>
<mi>arg</mi>
<munder>
<mi>min</mi>
<mi>&theta;</mi>
</munder>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>W</mi>
</munderover>
<mo>|</mo>
<mo>|</mo>
<msub>
<mi>y</mi>
<mi>i</mi>
</msub>
<mo>-</mo>
<msub>
<mi>&phi;</mi>
<mi>i</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>I</mi>
<mo>,</mo>
<mi>&theta;</mi>
<mo>)</mo>
</mrow>
<mo>|</mo>
<mo>|</mo>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>9</mn>
<mo>)</mo>
</mrow>
</mrow>
Here, | | | | represent Euclidean distance.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710572423.4A CN107451594B (en) | 2017-07-13 | 2017-07-13 | Multi-view gait classification method based on multiple regression |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710572423.4A CN107451594B (en) | 2017-07-13 | 2017-07-13 | Multi-view gait classification method based on multiple regression |
Publications (2)
Publication Number | Publication Date |
---|---|
CN107451594A true CN107451594A (en) | 2017-12-08 |
CN107451594B CN107451594B (en) | 2020-06-02 |
Family
ID=60488431
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710572423.4A Active CN107451594B (en) | 2017-07-13 | 2017-07-13 | Multi-view gait classification method based on multiple regression |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107451594B (en) |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109711387A (en) * | 2019-01-11 | 2019-05-03 | 哈尔滨工程大学 | A kind of gait image preprocess method based on multiclass energy diagram |
CN109784206A (en) * | 2018-12-25 | 2019-05-21 | 杭州电子科技大学 | A kind of gait recognition method based on the learning machine that transfinites |
CN110197130A (en) * | 2019-05-09 | 2019-09-03 | 广州番禺职业技术学院 | A kind of live pig abnormal gait detection device and system |
WO2020019171A1 (en) * | 2018-07-24 | 2020-01-30 | 中国科学院深圳先进技术研究院 | Identity authentication method and apparatus, electronic device, and storage medium |
CN110751084A (en) * | 2019-10-17 | 2020-02-04 | 中国计量大学 | Gait feature representation and extraction method based on line scanning statistical vector |
CN112712106A (en) * | 2020-12-07 | 2021-04-27 | 西安交通大学 | Mechanical equipment health state identification method based on multi-view confrontation self-encoder |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102122354A (en) * | 2011-03-15 | 2011-07-13 | 上海交通大学 | Adaptive characteristic block selection-based gait identification method |
CN102426645A (en) * | 2011-08-30 | 2012-04-25 | 北京航空航天大学 | Multi-view and multi-state gait recognition method |
CN104061907A (en) * | 2014-07-16 | 2014-09-24 | 中南大学 | Viewing-angle greatly-variable gait recognition method based on gait three-dimensional contour matching synthesis |
-
2017
- 2017-07-13 CN CN201710572423.4A patent/CN107451594B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102122354A (en) * | 2011-03-15 | 2011-07-13 | 上海交通大学 | Adaptive characteristic block selection-based gait identification method |
CN102426645A (en) * | 2011-08-30 | 2012-04-25 | 北京航空航天大学 | Multi-view and multi-state gait recognition method |
CN104061907A (en) * | 2014-07-16 | 2014-09-24 | 中南大学 | Viewing-angle greatly-variable gait recognition method based on gait three-dimensional contour matching synthesis |
Non-Patent Citations (2)
Title |
---|
王竣 等: "特征融合的多视角步态识别研究", 《中国计量大学学报》 * |
陈夏辉: "步态识别的若干关键技术研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2020019171A1 (en) * | 2018-07-24 | 2020-01-30 | 中国科学院深圳先进技术研究院 | Identity authentication method and apparatus, electronic device, and storage medium |
CN109784206A (en) * | 2018-12-25 | 2019-05-21 | 杭州电子科技大学 | A kind of gait recognition method based on the learning machine that transfinites |
CN109711387A (en) * | 2019-01-11 | 2019-05-03 | 哈尔滨工程大学 | A kind of gait image preprocess method based on multiclass energy diagram |
CN109711387B (en) * | 2019-01-11 | 2023-04-18 | 哈尔滨工程大学 | Gait image preprocessing method based on multi-class energy maps |
CN110197130A (en) * | 2019-05-09 | 2019-09-03 | 广州番禺职业技术学院 | A kind of live pig abnormal gait detection device and system |
CN110751084A (en) * | 2019-10-17 | 2020-02-04 | 中国计量大学 | Gait feature representation and extraction method based on line scanning statistical vector |
CN110751084B (en) * | 2019-10-17 | 2023-04-18 | 中国计量大学 | Gait feature representation and extraction method based on line scanning statistical vector |
CN112712106A (en) * | 2020-12-07 | 2021-04-27 | 西安交通大学 | Mechanical equipment health state identification method based on multi-view confrontation self-encoder |
CN112712106B (en) * | 2020-12-07 | 2022-12-09 | 西安交通大学 | Mechanical equipment health state identification method based on multi-view confrontation self-encoder |
Also Published As
Publication number | Publication date |
---|---|
CN107451594B (en) | 2020-06-02 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107451594A (en) | A kind of various visual angles Approach for Gait Classification based on multiple regression | |
CN105488534B (en) | Traffic scene deep analysis method, apparatus and system | |
CN105975931B (en) | A kind of convolutional neural networks face identification method based on multiple dimensioned pond | |
CN108304826A (en) | Facial expression recognizing method based on convolutional neural networks | |
CN106650806B (en) | A kind of cooperating type depth net model methodology for pedestrian detection | |
CN106407986B (en) | A kind of identification method of image target of synthetic aperture radar based on depth model | |
CN103942577B (en) | Based on the personal identification method for establishing sample database and composite character certainly in video monitoring | |
CN104217214B (en) | RGB D personage's Activity recognition methods based on configurable convolutional neural networks | |
CN105095833B (en) | For the network establishing method of recognition of face, recognition methods and system | |
CN110929736B (en) | Multi-feature cascading RGB-D significance target detection method | |
CN107506722A (en) | One kind is based on depth sparse convolution neutral net face emotion identification method | |
CN107229914B (en) | Handwritten digit recognition method based on deep Q learning strategy | |
CN107818302A (en) | Non-rigid multiple dimensioned object detecting method based on convolutional neural networks | |
CN105426919A (en) | Significant guidance and unsupervised feature learning based image classification method | |
CN106991408A (en) | The generation method and method for detecting human face of a kind of candidate frame generation network | |
CN106096535A (en) | A kind of face verification method based on bilinearity associating CNN | |
CN102509286B (en) | Target region sketching method for medical image | |
CN107657204A (en) | The construction method and facial expression recognizing method and system of deep layer network model | |
CN108256571A (en) | A kind of Chinese meal food recognition methods based on convolutional neural networks | |
CN109558902A (en) | A kind of fast target detection method | |
CN108053398A (en) | A kind of melanoma automatic testing method of semi-supervised feature learning | |
CN104103093B (en) | Three-dimensional grid semantic marking method based on deep convolution neural network | |
CN108122003A (en) | A kind of Weak target recognition methods based on deep neural network | |
CN108664994A (en) | A kind of remote sensing image processing model construction system and method | |
CN104050680B (en) | Based on iteration self-organizing and the image partition method of multi-agent genetic clustering algorithm |
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 |