CN110287783A - A kind of video monitoring image human figure identification method - Google Patents
A kind of video monitoring image human figure identification method Download PDFInfo
- Publication number
- CN110287783A CN110287783A CN201910415722.6A CN201910415722A CN110287783A CN 110287783 A CN110287783 A CN 110287783A CN 201910415722 A CN201910415722 A CN 201910415722A CN 110287783 A CN110287783 A CN 110287783A
- Authority
- CN
- China
- Prior art keywords
- image
- profile
- point
- curvature
- target
- 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
- 238000000034 method Methods 0.000 title claims abstract description 27
- 238000012544 monitoring process Methods 0.000 title claims abstract description 17
- 239000011159 matrix material Substances 0.000 claims abstract description 18
- 238000001914 filtration Methods 0.000 claims abstract description 11
- 230000009467 reduction Effects 0.000 claims abstract description 11
- 238000012545 processing Methods 0.000 claims abstract description 9
- 238000001514 detection method Methods 0.000 claims abstract description 7
- 238000012549 training Methods 0.000 claims abstract description 6
- 238000010586 diagram Methods 0.000 claims description 14
- 230000006870 function Effects 0.000 claims description 6
- 238000004364 calculation method Methods 0.000 claims description 4
- 230000007246 mechanism Effects 0.000 claims description 4
- 230000004069 differentiation Effects 0.000 claims description 3
- 229910017435 S2 In Inorganic materials 0.000 claims 1
- 238000012512 characterization method Methods 0.000 abstract description 2
- 238000012805 post-processing Methods 0.000 abstract description 2
- 230000008901 benefit Effects 0.000 description 2
- 230000008676 import Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000010191 image analysis Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000008569 process Effects 0.000 description 1
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
- G06V10/20—Image preprocessing
- G06V10/30—Noise filtering
-
- 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
- G06V10/46—Descriptors for shape, contour or point-related descriptors, e.g. scale invariant feature transform [SIFT] or bags of words [BoW]; Salient regional features
- G06V10/469—Contour-based spatial representations, e.g. vector-coding
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/40—Scenes; Scene-specific elements in video content
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
Abstract
The invention discloses a kind of video monitoring image human figure identification methods, include the following steps, S1: gaussian filtering noise reduction, smoothed image;S2: Three image difference handles video sequence;S3: plavini expands detection processing spatial dimension;S4: contours extract is carried out using Freeman chain code mode;S5: the curvature value of each point on profile is calculated;S6: curvature distribution curve and curvature space distribution matrix are obtained;S7: building learning model;S8: targets threshold matching differentiates humanoid profile;After the present invention carries out gaussian filtering and inter-frame difference to target, expand target post-processing range using plavini, the characterization quality of target can be effectively improved, discriminate whether it is humanoid target by the way of training study in conjunction with Learning model later, it is long to avoid tradition recognition time with high accuracy, the lower disadvantage of the short ratio of precision of recognition time.
Description
Technical field
The present invention relates to computerized algorithm technical field, specially a kind of video monitoring image human figure identification method.
Background technique
Carrying out the method that humanoid target identification mainly uses to monitor video at present has very much, classification three categories, and one
Class has unique advantage in feature description, for example describes to identify that humanoid target mentions using combinations such as HOG feature, LBP features
High humanoid goal description ability, the second class mainly have its advantage on classifier, such as using classical point of adaboost, SVM etc.
Class device carries out the Classification and Identification of humanoid target, and third class is mainly the knowledge for constructing neural network model to carry out humanoid target
Not, which robustness is preferable, but optimization can only lean on data constantly to train iteration.
Existing humanoid target identification method all has certain problems, and recognition time with high accuracy is long, when identification
Between short precision it is relatively low;And there is presently no a standardized Human detection algorithm, all kinds of relevant algorithm steps all compare
It is more.
Summary of the invention
The technical problem to be solved by the present invention is to existing humanoid target identification methods all to have certain problems, and precision is high
Recognition time it is long, the short precision of recognition time is relatively low;And there is presently no a standardized Human detection algorithm,
All kinds of relevant algorithm steps are all relatively more, a kind of video monitoring image human figure identification method provided, to solve the above problems.
In order to solve the above-mentioned technical problems, the present invention provides the following technical solutions:
The present invention provides a kind of video monitoring image human figure identification method, includes the following steps,
S1: gaussian filtering noise reduction, smoothed image;
S2: Three image difference handles video sequence;
S3: plavini expands detection processing spatial dimension;
S4: contours extract is carried out using Freeman chain code mode;
S5: the curvature value of each point on profile is calculated;
S6: curvature distribution curve and curvature space distribution matrix are obtained;
S7: building learning model;
S8: targets threshold matching differentiates humanoid profile.
As a preferred technical solution of the present invention, in the step S1, the used function of gaussian filtering are as follows:The Gaussian kernel of discrete distribution is constructed simultaneously;The central element of Gaussian kernel is moved on the image, it is high
The central element of this core is matched with image slices vegetarian refreshments, later using the pixel value of image as weight, multiplied by Gaussian kernel as final
Value;Each point on image is operated, the noise reduction for completing image is smooth.
As a preferred technical solution of the present invention, in the step S2, the Three image difference formula are as follows: D'n(x,
Y)=| Fn+1(x,y)-Fn(x,y)|∩|Fn(x,y)-Fn-1(x,y)|;The (n+1)th frame, n-th frame and (n-1)th in note video sequence
The image of frame is respectively Fn+1, Fn, Fn-1, and the gray value of three frame corresponding pixel points is denoted as Fn+1 (x, y), Fn (x, y), Fn-1
(x, y) obtains difference image Dn+1, Dn according to formula, carries out and operate to difference image, obtain final image, carry out threshold value
Processing, to obtain destination image data.
As a preferred technical solution of the present invention, in the step S3, the plavini is used at an expansion core
The image obtained in reason step S2, the expansion core formula are as follows:
As a preferred technical solution of the present invention, in the step S4, movement mesh is calculated using Freeman chain code
The relationship between quality for marking adjacent position between each point, using 7*7 block of pixels as basic unit, since the key point in the upper left corner, along
Objective contour calculates the direction vector of each block of pixels, after calculating, obtains the coordinate and position of one group of record start point
The Freeman chain code sequence point for setting direction, the profile of target can be constituted by connecting these points.
As a preferred technical solution of the present invention, in the step S5, the calculation formula of the curvature are as follows:With the calculated point center of circle, all the points are formed point by radius r, share n point, then using minimum two
Multiplication carries out curve fitting to obtain curvilinear equation, to calculate the curvature of the point.
As a preferred technical solution of the present invention, the step of curvature estimation, is as follows,
If setting polynomial fitting as y=a0+a1x+...+akxk;
Each point is calculated to the sum of the distance of curve, is
Partial derivative is sought on the right of peer-to-peer, obtains following matrix:
Matrix reduction is obtained:
Matrix is coefficient matrix, carries out once differentiation to polynomial function and obtains derivative: y'=a1+...+akxk-1;
Finally profile point is substituted into derivative and obtains its curvature value k.
As a preferred technical solution of the present invention, in the step S6, select profile highest point for starting point, according to
Clock-wise fashion obtains entire profile with target image laterally for axis of ordinates with the longitudinal for axis of abscissas of target image
Coordinate system;Centered on profile diagram central point, using the ordinate of target image as central axes, left half side profile diagram and the right side are obtained
Half side profile diagram obtains upper half side profile figure and lower half side profile figure using the abscissa of target image as central axes, by above five
The curvature and coordinate of a profile are distributed to form five tables of data sequences.
As a preferred technical solution of the present invention, in the step S7, Learning learning model mechanism is introduced, it is right
The humanoid profile image of label forms training set, is pre-processed the humanoid profile characteristic data set marked.
As a preferred technical solution of the present invention, in the step S8, after unlabelled profile diagram is handled
Characteristic sequence is obtained, the humanoid profile characteristic which imports label is concentrated, itself and the people in step S7 are calculated
The relevance values of shape contour feature data set obtain a relevance values sequence, by minimum threshold method, have one in the sequence
Value is greater than threshold value, is determined as humanoid profile, none value of the sequence is greater than threshold value, then is determined as it not being humanoid profile.
The beneficial effects obtained by the present invention are as follows being: the present invention mainly proposes a kind of based on the humanoid target of Computer Automatic Recognition
Algorithm can quickly identify humanoid target and extract its data, by carrying out gaussian filtering and frame-to-frame differences to target
After point, expands target post-processing range using plavini, the characterization quality of target can be effectively improved, obtain contour curvature later
Related data discriminates whether it is humanoid target by the way of training study in conjunction with Learning model.
Detailed description of the invention
Attached drawing is used to provide further understanding of the present invention, and constitutes part of specification, with reality of the invention
It applies example to be used to explain the present invention together, not be construed as limiting the invention.
In the accompanying drawings:
Fig. 1 is overall step schematic diagram of the present invention;
Fig. 2 is Three image difference schematic diagram of the present invention.
Specific embodiment
Hereinafter, preferred embodiments of the present invention will be described with reference to the accompanying drawings, it should be understood that preferred reality described herein
Apply example only for the purpose of illustrating and explaining the present invention and is not intended to limit the present invention.
Embodiment: as shown in Figs. 1-2, the present invention provides a kind of video monitoring image human figure identification method, including walks as follows
Suddenly,
S1: gaussian filtering noise reduction, smoothed image;
S2: Three image difference handles video sequence;
S3: plavini expands detection processing spatial dimension;
S4: contours extract is carried out using Freeman chain code mode;
S5: the curvature value of each point on profile is calculated;
S6: curvature distribution curve and curvature space distribution matrix are obtained;
S7: building learning model;
S8: targets threshold matching differentiates humanoid profile.
Further, in the step S1, the used function of gaussian filtering are as follows:Structure simultaneously
Build the Gaussian kernel of discrete distribution;The central element of Gaussian kernel, the central element and image slices vegetarian refreshments of Gaussian kernel are moved on the image
Matching, later using the pixel value of image as weight, multiplied by Gaussian kernel as end value;Each point on image is grasped
Make, the noise reduction for completing image is smooth.
Further, in the step S2, the Three image difference formula are as follows: D'n(x, y)=| Fn+1(x,y)-Fn(x,y)
|∩|Fn(x,y)-Fn-1(x,y)|;The image of the (n+1)th frame, n-th frame and the (n-1)th frame in note video sequence be respectively Fn+1,
Fn, Fn-1, the gray value of three frame corresponding pixel points are denoted as Fn+1 (x, y), Fn (x, y), Fn-1 (x, y), and it is poor to be obtained according to formula
Partial image Dn+1, Dn carry out difference image and operate, and obtain final image, threshold process are carried out, to obtain target image
Data.
Further, in the step S3, the plavini expands the image obtained in core processing step S2 using one,
The expansion core formula are as follows:
Further, in the step S4, adjacent position between moving target each point is calculated using Freeman chain code
Relationship between quality since the key point in the upper left corner, calculates each along objective contour using 7*7 block of pixels as basic unit
The direction vector of block of pixels obtains the coordinate of one group of record start point and the Freeman chain code sequence of locality after calculating
Column point, the profile of target can be constituted by connecting these points.
Further, in the step S5, the calculation formula of the curvature are as follows:With the calculated point center of circle, partly
Diameter is r, and all the points are formed point, shares n point, then carries out curve fitting to obtain curve side using least square method
Journey, to calculate the curvature of the point.
Further, the step of curvature estimation is as follows,
If setting polynomial fitting as y=a0+a1x+...+akxk;
Each point is calculated to the sum of the distance of curve, is
Partial derivative is sought on the right of peer-to-peer, obtains following matrix:
Matrix reduction is obtained:
Matrix is coefficient matrix, carries out once differentiation to polynomial function and obtains derivative: y'=a1+...+akxk-1;
Finally profile point is substituted into derivative and obtains its curvature value k.
Further, in the step S6, select profile highest point for starting point, according to clock-wise fashion, with target figure
The longitudinal direction of picture is axis of abscissas, with target image laterally for axis of ordinates, obtains the coordinate system of entire profile;In profile diagram
Centered on heart point, using the ordinate of target image as central axes, left half side profile diagram and right half side profile diagram are obtained, with target figure
The abscissa of picture is central axes, upper half side profile figure and lower half side profile figure is obtained, by the curvature and coordinate of above five profiles
Distribution forms five tables of data sequences.
Further, in the step S7, Learning learning model mechanism is introduced, to the humanoid profile image shape of label
At training set, the humanoid profile characteristic data set marked is pre-processed.
Further, in the step S8, characteristic sequence is obtained after unlabelled profile diagram is handled, by this
The humanoid profile characteristic that sequence imports label is concentrated, and it is related to the humanoid profile characteristic data set in step S7 to calculate it
Property value, obtain a relevance values sequence, by minimum threshold method, have in the sequence value be greater than threshold value, be determined as humanoid
Profile, none value of the sequence are greater than threshold value, are then determined as it not being humanoid profile.
It is specific: gaussian filtering noise reduction mainly is carried out to monitor video image in step sl, noise signal is filtered out,
Smoothed image simultaneously, is conducive to subsequent image analysis processing, after carrying out gaussian filtering;The description of moving target in order to obtain
Information and spatial positional information in step s 2 handle video sequence using Three image difference, obtain moving target
Description information and spatial positional information;Later, in step s3, in order to maximize reservation target data, expanded using plavini
Detection processing spatial dimension;Step S4 carries out contours extract using Freeman chain code mode, obtains the profile information of target;It is logical
Cross the curvature value that step S5 calculates each point on profile;To form the statistical matrix of spatial position and curvature value, song is obtained
Rate distribution curve and curvature space distribution matrix, i.e. step S6;In the step s 7, Learning learning model mechanism is introduced, it is right
The humanoid profile image of label forms training set, is pre-processed the humanoid profile characteristic data set marked;Final step
In S8, correlation calculations are done, finally determines to whether there is humanoid target in video image and mention using targets threshold matching way
It takes out.
Finally, it should be noted that these are only the preferred embodiment of the present invention, it is not intended to restrict the invention, although
Present invention has been described in detail with reference to the aforementioned embodiments, for those skilled in the art, still can be right
Technical solution documented by foregoing embodiments is modified or equivalent replacement of some of the technical features.It is all
Within the spirit and principles in the present invention, any modification, equivalent replacement, improvement and so on should be included in protection of the invention
Within the scope of.
Claims (10)
1. a kind of video monitoring image human figure identification method, which is characterized in that include the following steps,
S1: gaussian filtering noise reduction, smoothed image;
S2: Three image difference handles video sequence;
S3: plavini expands detection processing spatial dimension;
S4: contours extract is carried out using Freeman chain code mode;
S5: the curvature value of each point on profile is calculated;
S6: curvature distribution curve and curvature space distribution matrix are obtained;
S7: building learning model;
S8: targets threshold matching differentiates humanoid profile.
2. a kind of video monitoring image human figure identification method according to claim 1, which is characterized in that the step S1
In, the used function of gaussian filtering are as follows:The Gaussian kernel of discrete distribution is constructed simultaneously;On the image
The central element of mobile Gaussian kernel, the central element of Gaussian kernel are matched with image slices vegetarian refreshments, later using the pixel value of image as
Weight, multiplied by Gaussian kernel as end value;Each point on image is operated, the noise reduction for completing image is smooth.
3. a kind of video monitoring image human figure identification method according to claim 1, which is characterized in that the step S2
In, the Three image difference formula are as follows: D 'n(x, y)=| Fn+1(x,y)-Fn(x,y)|∩|Fn(x,y)-Fn-1(x,y)|;Note view
The image of the (n+1)th frame, n-th frame and the (n-1)th frame in frequency sequence is respectively Fn+1, Fn, Fn-1, the gray scale of three frame corresponding pixel points
Value is denoted as Fn+1 (x, y), Fn (x, y), Fn-1 (x, y), obtains difference image Dn+1, Dn according to formula, carries out to difference image
With operation, final image is obtained.
4. a kind of video monitoring image human figure identification method according to claim 1, which is characterized in that the step S3
In, the plavini is using the image obtained in an expansion core processing step S2, the expansion core formula are as follows:
5. a kind of video monitoring image human figure identification method according to claim 1, which is characterized in that the step S4
In, the relationship between quality of adjacent position between moving target each point is calculated using Freeman chain code, is basic with 7*7 block of pixels
Unit calculates the direction vector of each block of pixels along objective contour since the key point in the upper left corner, by calculating
Afterwards, as soon as obtaining the Freeman chain code sequence point of group coordinate of record start point and locality, mesh can be constituted by connecting these points
Target profile.
6. a kind of video monitoring image human figure identification method according to claim 1, which is characterized in that the step S5
In, the calculation formula of the curvature are as follows:With the calculated point center of circle, all the points are formed point by radius r,
N point is shared, then carries out curve fitting to obtain curvilinear equation using least square method, to calculate the curvature of the point.
7. a kind of video monitoring image human figure identification method according to claim 6, which is characterized in that the curvature estimation
The step of it is as follows,
If setting polynomial fitting as y=a0+a1x+...+akxk;
Each point is calculated to the sum of the distance of curve, is
Partial derivative is sought on the right of peer-to-peer, obtains following matrix:
Matrix reduction is obtained:
Matrix is coefficient matrix, carries out once differentiation to polynomial function and obtains derivative: y'=a1+...+akxk-1;
Finally profile point is substituted into derivative and obtains its curvature value k.
8. a kind of video monitoring image human figure identification method according to claim 1, which is characterized in that the step S6
In, select profile highest point for starting point, according to clock-wise fashion, with the longitudinal for axis of abscissas of target image, with target figure
The transverse direction of picture is axis of ordinates, obtains the coordinate system of entire profile;Centered on profile diagram central point, with the vertical seat of target image
Central axes are designated as, left half side profile diagram and right half side profile diagram is obtained using the abscissa of target image as central axes and obtains upper half
Side profile figure and lower half side profile figure are distributed the curvature of above five profiles and coordinate to form five tables of data sequences.
9. a kind of video monitoring image human figure identification method according to claim 1, which is characterized in that the step S7
In, Learning learning model mechanism is introduced, training set is formed to the humanoid profile image of label, is pre-processed and is marked
Humanoid profile characteristic data set.
10. a kind of video monitoring image human figure identification method according to claim 1, which is characterized in that the step S8
In, characteristic sequence is obtained after unlabelled profile diagram is handled, which is imported to the humanoid profile feature of label
In data set, the relevance values of itself and the humanoid profile characteristic data set in step S7 are calculated, obtain a relevance values sequence,
By minimum threshold method, there is a value to be greater than threshold value in the sequence, be determined as humanoid profile, none value of the sequence is greater than threshold
Value, then be determined as it not being humanoid profile.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910415722.6A CN110287783A (en) | 2019-05-18 | 2019-05-18 | A kind of video monitoring image human figure identification method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910415722.6A CN110287783A (en) | 2019-05-18 | 2019-05-18 | A kind of video monitoring image human figure identification method |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110287783A true CN110287783A (en) | 2019-09-27 |
Family
ID=68002163
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910415722.6A Pending CN110287783A (en) | 2019-05-18 | 2019-05-18 | A kind of video monitoring image human figure identification method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110287783A (en) |
Citations (15)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US7016539B1 (en) * | 1998-07-13 | 2006-03-21 | Cognex Corporation | Method for fast, robust, multi-dimensional pattern recognition |
US20070047790A1 (en) * | 2005-08-30 | 2007-03-01 | Agfa-Gevaert N.V. | Method of Segmenting Anatomic Entities in Digital Medical Images |
US20090092301A1 (en) * | 2007-10-03 | 2009-04-09 | Siemens Medical Solutions Usa, Inc. | System and Method for Organ Segmentation Using Surface Patch Classification in 2D and 3D Images |
CN101763504A (en) * | 2009-12-30 | 2010-06-30 | 邱怀骏 | Human head identification method under complex scene |
CN102722721A (en) * | 2012-05-25 | 2012-10-10 | 山东大学 | Human falling detection method based on machine vision |
US20140212048A1 (en) * | 2012-02-27 | 2014-07-31 | Aselsan Elektornik Sanayi Ve Ticaret Anonim Sirket | System and Method for Identifying Scale Invariant Features of Object Outlines on Images |
US20170039761A1 (en) * | 2014-05-14 | 2017-02-09 | Huawei Technologies Co., Ltd. | Image Processing Method And Apparatus |
CN106408846A (en) * | 2016-11-29 | 2017-02-15 | 周川 | Image fire hazard detection method based on video monitoring platform |
CN106485185A (en) * | 2015-08-27 | 2017-03-08 | 无锡林之盛科技有限公司 | A kind of detection algorithm is fallen down based on the old man of boundary chain code |
CN106846359A (en) * | 2017-01-17 | 2017-06-13 | 湖南优象科技有限公司 | Moving target method for quick based on video sequence |
CN107273815A (en) * | 2017-05-24 | 2017-10-20 | 中国农业大学 | A kind of individual behavior recognition methods and system |
CN107590440A (en) * | 2017-08-21 | 2018-01-16 | 南京邮电大学 | The method and system of Human detection under a kind of Intelligent household scene |
CN108154066A (en) * | 2016-12-02 | 2018-06-12 | 中国科学院沈阳自动化研究所 | A kind of Three-dimensional target recognition method based on curvature feature recurrent neural network |
CN109409208A (en) * | 2018-09-10 | 2019-03-01 | 东南大学 | A kind of vehicle characteristics extraction and matching process based on video |
CN109711284A (en) * | 2018-12-11 | 2019-05-03 | 江苏博墨教育科技有限公司 | A kind of test answer sheet system intelligent recognition analysis method |
-
2019
- 2019-05-18 CN CN201910415722.6A patent/CN110287783A/en active Pending
Patent Citations (15)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US7016539B1 (en) * | 1998-07-13 | 2006-03-21 | Cognex Corporation | Method for fast, robust, multi-dimensional pattern recognition |
US20070047790A1 (en) * | 2005-08-30 | 2007-03-01 | Agfa-Gevaert N.V. | Method of Segmenting Anatomic Entities in Digital Medical Images |
US20090092301A1 (en) * | 2007-10-03 | 2009-04-09 | Siemens Medical Solutions Usa, Inc. | System and Method for Organ Segmentation Using Surface Patch Classification in 2D and 3D Images |
CN101763504A (en) * | 2009-12-30 | 2010-06-30 | 邱怀骏 | Human head identification method under complex scene |
US20140212048A1 (en) * | 2012-02-27 | 2014-07-31 | Aselsan Elektornik Sanayi Ve Ticaret Anonim Sirket | System and Method for Identifying Scale Invariant Features of Object Outlines on Images |
CN102722721A (en) * | 2012-05-25 | 2012-10-10 | 山东大学 | Human falling detection method based on machine vision |
US20170039761A1 (en) * | 2014-05-14 | 2017-02-09 | Huawei Technologies Co., Ltd. | Image Processing Method And Apparatus |
CN106485185A (en) * | 2015-08-27 | 2017-03-08 | 无锡林之盛科技有限公司 | A kind of detection algorithm is fallen down based on the old man of boundary chain code |
CN106408846A (en) * | 2016-11-29 | 2017-02-15 | 周川 | Image fire hazard detection method based on video monitoring platform |
CN108154066A (en) * | 2016-12-02 | 2018-06-12 | 中国科学院沈阳自动化研究所 | A kind of Three-dimensional target recognition method based on curvature feature recurrent neural network |
CN106846359A (en) * | 2017-01-17 | 2017-06-13 | 湖南优象科技有限公司 | Moving target method for quick based on video sequence |
CN107273815A (en) * | 2017-05-24 | 2017-10-20 | 中国农业大学 | A kind of individual behavior recognition methods and system |
CN107590440A (en) * | 2017-08-21 | 2018-01-16 | 南京邮电大学 | The method and system of Human detection under a kind of Intelligent household scene |
CN109409208A (en) * | 2018-09-10 | 2019-03-01 | 东南大学 | A kind of vehicle characteristics extraction and matching process based on video |
CN109711284A (en) * | 2018-12-11 | 2019-05-03 | 江苏博墨教育科技有限公司 | A kind of test answer sheet system intelligent recognition analysis method |
Non-Patent Citations (2)
Title |
---|
RONGFENG LI,: "A method for countering dense false target jamming based on correlation sample selection", 《DBLP》, 31 December 2018 (2018-12-31) * |
傅春燕;李平;文玉梅;袁海军;叶波;: "一种采用高斯模型的步态轮廓分割算法", 传感技术学报, no. 07, 15 July 2008 (2008-07-15) * |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109460702B (en) | Passenger abnormal behavior identification method based on human body skeleton sequence | |
CN105930767B (en) | A kind of action identification method based on human skeleton | |
CN108921107B (en) | Pedestrian re-identification method based on sequencing loss and Simese network | |
CN107403154B (en) | Gait recognition method based on dynamic vision sensor | |
CN110084228A (en) | A kind of hazardous act automatic identifying method based on double-current convolutional neural networks | |
CN103971386B (en) | A kind of foreground detection method under dynamic background scene | |
CN109544613A (en) | A kind of binocular solid matching process and system based on the study of dense network depth | |
CN110084215A (en) | A kind of pedestrian of the twin network model of binaryzation triple recognition methods and system again | |
CN101833654B (en) | Sparse representation face identification method based on constrained sampling | |
CN107229920B (en) | Behavior identification method based on integration depth typical time warping and related correction | |
CN107833239B (en) | Optimization matching target tracking method based on weighting model constraint | |
CN111199556B (en) | Indoor pedestrian detection and tracking method based on camera | |
CN111339988A (en) | Video face recognition method based on dynamic interval loss function and probability characteristic | |
CN107451594B (en) | Multi-view gait classification method based on multiple regression | |
JP7422456B2 (en) | Image processing device, image processing method and program | |
CN104599291B (en) | Infrared motion target detection method based on structural similarity and significance analysis | |
CN107863153A (en) | A kind of human health characteristic modeling measuring method and platform based on intelligent big data | |
CN104038799A (en) | Three-dimensional television-oriented gesture manipulation method | |
CN108921064B (en) | Pedestrian re-identification method based on multi-feature fusion | |
CN110458235A (en) | Movement posture similarity comparison method in a kind of video | |
CN104217211B (en) | Multi-visual-angle gait recognition method based on optimal discrimination coupling projection | |
Shah et al. | Multi-camera handoff for person re-identification | |
CN105869153A (en) | Non-rigid face image registering method integrated with related block information | |
CN107967455A (en) | A kind of transparent learning method of intelligent human-body multidimensional physical feature big data and system | |
CN113327269A (en) | Unmarked cervical vertebra movement detection method |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination |