CN106295532A - A kind of human motion recognition method in video image - Google Patents
A kind of human motion recognition method in video image Download PDFInfo
- Publication number
- CN106295532A CN106295532A CN201610621491.0A CN201610621491A CN106295532A CN 106295532 A CN106295532 A CN 106295532A CN 201610621491 A CN201610621491 A CN 201610621491A CN 106295532 A CN106295532 A CN 106295532A
- Authority
- CN
- China
- Prior art keywords
- rectangular histogram
- obtains
- profile energy
- classification
- histogram
- 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
- 238000000034 method Methods 0.000 title claims abstract description 78
- 238000012549 training Methods 0.000 claims abstract description 15
- 230000008859 change Effects 0.000 claims abstract description 11
- 238000012216 screening Methods 0.000 claims abstract description 6
- 230000001149 cognitive effect Effects 0.000 claims abstract description 4
- 230000009471 action Effects 0.000 claims description 16
- 230000008569 process Effects 0.000 claims description 14
- 239000012535 impurity Substances 0.000 claims description 12
- 238000003064 k means clustering Methods 0.000 claims description 5
- 238000013507 mapping Methods 0.000 claims description 3
- 238000011410 subtraction method Methods 0.000 claims description 3
- 230000011218 segmentation Effects 0.000 abstract description 4
- 238000012360 testing method Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 3
- 230000004044 response Effects 0.000 description 3
- 230000008901 benefit Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012544 monitoring process Methods 0.000 description 2
- 239000000654 additive Substances 0.000 description 1
- 230000000996 additive effect Effects 0.000 description 1
- 230000003542 behavioural effect Effects 0.000 description 1
- 239000000969 carrier Substances 0.000 description 1
- 238000002790 cross-validation Methods 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000003909 pattern recognition Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- 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
- G06V20/41—Higher-level, semantic clustering, classification or understanding of video scenes, e.g. detection, labelling or Markovian modelling of sport events or news items
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Multimedia (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Computational Linguistics (AREA)
- Software Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Human Computer Interaction (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses the human motion recognition method in a kind of video image, comprise the following steps: one, each frame input picture is carried out pretreatment and obtains foreground area, foreground area is carried out screening and obtains target area；Two, obtain objective contour according to target area；Three, it is thus achieved that the profile energy variation rectangular histogram of X and Y-direction；Four, profile energy variation rectangular histogram is normalized；Five, the training stage: the training set forming profile energy variation rectangular histogram carries out the classification of motion, obtains human body behavior model and gives weights；Six, at cognitive phase: the human body behavior model that the profile energy histogram of frame to be measured obtained with the training stage is mated, execution identification.The present invention obtains profile energy variation rectangular histogram by calculating the change of objective contour in consecutive frame, carries out unsupervised segmentation according to profile energy variation rectangular histogram, improves accuracy rate and robustness, ensure that real-time simultaneously.
Description
Technical field
The present invention relates to the human body recognition method in a kind of video image, belong to the technology neck of image procossing and pattern recognition
Territory.
Background technology
Along with obtaining video equipment and the fast development of broadband network, video is as the main carriers of information.Greatly
Most videos are all the activities of the people of record, so whether from safety, monitoring and entertain, or the angle of personal information storage
Degree, the research being identified the human action in video is just provided with highly important learning value and application prospect.From this
For in matter, Human bodys' response is exactly that the pedestrian target split is extracted feature interested, then to extracting
Characteristic carry out sort operation.At present, conventional Human bodys' response method can be divided into method based on template matching
And method of based on state space.Method based on template matching is the sequence image template with reference to human body behavior to be left in
In data base, afterwards testing image is mated with the reference sequences image deposited in data base, thus find similarity
The highest reference sequences image, and then determine human body behavior classification to be tested.Human bodys' response method based on template is multiple
Miscellaneous degree is relatively low, but does not accounts for human body behavior dynamic characteristic in the video sequence, and the most sensitive to noise jamming.
Method based on state space is by describing the feature of human motion, the basic poses of human body behavior when being made a state,
Being advanced between these states by certain probabilistic relation, wherein applying most is hidden Markov model.But human body row
There is also, for identifying, the difficult problem much needing to overcome at present, human body is non-rigid targets, and everyone does identical action
The property of there are differences, this just brings to the generality of Activity recognition and there is also similarity between difficulty, moreover some human action
And action is of a great variety, this is all the problem to be considered when designing Activity recognition method.
At present, intelligent monitoring is more and more higher for the requirement of real-time and accuracy, and traditional method is difficult to meet now
The demand of actual application.
Summary of the invention
It is an object of the invention to overcome deficiency of the prior art, it is provided that the human action in a kind of video image is known
Other method, solves tradition low based on model generalization ability in model matching method, and noise resisting ability is poor and empty based on state
Between the technical problem of action classification similarity in method.
For solving above-mentioned technical problem, the invention provides the human motion recognition method in a kind of video image, it is special
Levy and be, comprise the following steps:
Step one, carries out pretreatment to each frame input picture and obtains foreground area, foreground area is carried out screening and obtains mesh
Mark region；
Step 2, obtains objective contour according to target area；
Step 3, it is thus achieved that the profile energy variation rectangular histogram of X and Y-direction；
Step 4, is normalized profile energy variation rectangular histogram；
Step 5, the training stage: the training set forming profile energy variation rectangular histogram carries out the classification of motion, obtains human body
Behavior model also gives weights；
Step 6, at cognitive phase: the human body behavior mould obtained profile energy histogram and the training stage of frame to be measured
Type mates, execution identification.
Further, in described step one, pretreatment uses background subtraction method, screening to use minimum enclosed rectangle frame
Method.
Further, in described step 3, obtaining profile energy variation histogram method is:
31) the edge image I of adjacent two two field pictures is obtainededgeAnd Ilast_edge, use 10 × 10 windows by row traversal edge
Image Iedge；
32) during traversal, when there is edge pixel in window, at previous frame image Ilast_edgeIn same area find
The edge pixel that Euclidean distance is minimum therewith matches, and the size of Euclidean distance is changed as this point edge pixel energy
Value；
33) after having traveled through, will row number as histogrammic abscissa, energy change value corresponding to each column is as rectangular histogram
Vertical coordinate, obtain profile energy variation rectangular histogram.
Further, in described step 4, normalized process is, first rectangular histogram vertical coordinate is normalized place
Reason so that it is value is between 0 to 1, is then the rectangular histogram of fixed size by Histogram Mapping a to abscissa.
Further, in described step 5, sorting technique is:
51) utilize k-means clustering method to obtain cluster barycenter collection, behavior is carried out big class division, each division obtained
Classification Ci, wherein, 1≤i≤n, n are behavior classification numbers；
52) utilize Euclidean distance to each CiContrast two-by-two, obtain Geordie impurity level Gi, Geordie impurity level GiAs class
Other CiWeights.
Further, in described step 6, the detailed process in identification of behavior to be measured is:
61) for behavior S to be measuredq={ K1,K2,K3,.......,KlCarry out step one to three process and obtain its profile energy
Amount change rectangular histogram, it is judged that image KtThe Euclidean distance of barycenter of rectangular histogram and each classification, choose Euclidean distance minimum
Classification is as image KtAffiliated classification Ci, wherein, 1≤t≤l；
62) by SqThe probability belonging to each swooping template action behavior is set to Aq={ A1,A2,A3,......,An, wherein may
Property AiIt is according to Geordie impurity level GiTo CiIt is optimized and obtains, Ai=Gi/Ci；
63) according to each two field picture generic Ai, select maximum AmaxSo that it is determined that SqType of action.
Compared with prior art, the present invention is reached to provide the benefit that: the present invention calculates adjacent by Euclidean distance
In frame, the change of objective contour obtains profile energy variation rectangular histogram, utilizes k-means clustering method to obtain each two field picture
Profile energy variation rectangular histogram carries out unsupervised segmentation, gives weights by Geordie impurity level to classification results, improves accurately
Rate and robustness, ensure that real-time simultaneously, solves tradition low based on model generalization ability in model matching method, anti-noise
Sound ability and based on the problem of action classification similarity in state-space method.
Accompanying drawing explanation
Fig. 1 is the flow chart of the inventive method.
Fig. 2 is the image of boxing behavior in embodiment of the present invention KTH data base.
Fig. 3 is the image of handclapping behavior in embodiment of the present invention KTH data base.
Fig. 4 is the image of handwaving behavior in embodiment of the present invention KTH data base.
Fig. 5 is the image of jogging behavior in embodiment of the present invention KTH data base.
Fig. 6 is the image of running behavior in embodiment of the present invention KTH data base.
Fig. 7 is the image of walking behavior in embodiment of the present invention KTH data base.
Detailed description of the invention
The invention will be further described below in conjunction with the accompanying drawings.Following example are only used for clearly illustrating the present invention
Technical scheme, and can not limit the scope of the invention with this.
As it is shown in figure 1, the human motion recognition method in a kind of video image of the present invention, it is characterized in that, including following
Step:
Step one, carries out pretreatment to each frame input picture and obtains foreground area, foreground area is carried out screening and obtains mesh
Mark region；
Behavioral training collection S={S1,S2,S3,......,Sn(n is behavior classification number), wherein, behavior Si(wherein, 1≤i
≤ n), behavior Si={ K1,K2,K3,.......,Km(m is number of image frames), Kj(wherein, 1≤j≤m) is composition behavior Si's
Each two field picture, uses background subtraction method to obtain foreground area each frame input picture in one video, and its detailed process sees existing
There is technology, then comprise foreground area by minimum enclosed rectangle frame, thus determine whether human body target region, filter out mesh
Mark region.
Step 2, obtains objective contour according to target area；
The method obtaining objective contour is, first uses 2D gaussian filtering template to be filtered input picture, then utilizes
Canny operator extracts human body attitude two-value profile frame by frame, then by Sobel operator, each edge pixel in image is calculated it
The size and Orientation of gradient.
Step 3, it is thus achieved that the profile energy variation rectangular histogram of X and Y-direction；
Calculate the change of objective contour in consecutive frame by Euclidean distance and obtain profile energy variation rectangular histogram, obtain wheel
The wide histogrammic detailed process of energy variation is:
31) the edge image I of adjacent two two field pictures is obtainededgeAnd Ilast_edge, use 10 × 10 windows by row traversal edge
Image Iedge；
32) during traversal, when there is edge pixel in window, at previous frame image Ilast_edgeIn same area find
The edge pixel that Euclidean distance is minimum therewith matches, and the size of Euclidean distance is changed as this point edge pixel energy
Value；
33) after having traveled through, will row number as histogrammic abscissa, energy change value corresponding to each column is as rectangular histogram
Vertical coordinate, obtain profile energy variation rectangular histogram.
Step 4, is normalized profile energy variation rectangular histogram；
Normalized process is, is first normalized rectangular histogram vertical coordinate so that it is value is between 0 to 1, so
After be the rectangular histogram of fixed size by Histogram Mapping a to abscissa.
Step 5, the training stage: the training set forming profile energy variation rectangular histogram carries out the classification of motion, obtains human body
Behavior model also gives weights；
The profile energy variation rectangular histogram utilizing k-means clustering method to obtain each two field picture carries out unsupervised segmentation,
The detailed process of classification is:
51) randomly choosing k object in the training set being made up of profile variations energy histogram, each object represents one
The barycenter of individual cluster；Wherein the value of k empirically chooses 3≤k≤n；
52) for remaining each object, according to the distance between this object and each cluster barycenter, it is assigned to therewith
In most like cluster；
53) the new barycenter of each cluster is calculated；
54) above-mentioned 51 are repeated)-53) process, until criterion function is assembled；
55) according to cluster barycenter collection R achieved aboven, behavior S is carried out big class division, each stroke obtained is categorized as Ci；
56) utilize Euclidean distance to CiContrast two-by-two, obtain Geordie impurity level Gi, Geordie impurity level GiAs classification
CiWeights；The process wherein obtaining Geordie impurity level sees prior art.
Step 6, at cognitive phase: the human body behavior mould obtained profile energy histogram and the training stage of frame to be measured
Type mates, execution identification.
The detailed process in identification of behavior to be measured is:
61) for behavior S to be measuredq={ K1,K2,K3,.......,KlCarry out step one to three process and obtain its profile energy
Amount change rectangular histogram, it is judged that image KtThe rectangular histogram of (wherein, 1≤t≤l) and the Euclidean distance of the barycenter of each classification, choose
The classification of Euclidean distance minimum is as image KtAffiliated classification Ci,
62) by SqThe probability belonging to each swooping template action behavior is set to Aq={ A1,A2,A3,......,An, wherein may
Property AiIt is according to Geordie impurity level GiTo CiIt is optimized and obtains, Ai=Gi/Ci；AiIt is worth the biggest, represents SqBelong to the i-th action classification
Probability the biggest, use ratio optimize, the discrimination between inhomogeneity can be improved；
63) according to probability A of each two field picture generici, select maximum AmaxSo that it is determined that SqType of action.
Embodiment one
The present invention use leaving-one method (assume there is N number of sample, using each sample as test sample, other N-1 sample
As training sample) method is carried out cross validation, test sample uses KTH human body behavior database, and this data base includes 6 classes
Behavior: boxing, jogging, running, boxing, handwaving, handclapping, is to be held by 25 different people
Row, respectively under four scenes (outdoor background, camera lens furthers and zooms out, video camera light exercise, room background), one has
599 sections of videos.Fig. 2 to Fig. 7 be respectively boxing in KTH data base, handclapping, handwaving, jogging,
The image of running and walking behavior.Prior art is carried out human action identification use method have Schindler,
Ahmad, Jhuang, Rodriguez and Mikolajczyk.Based on 6 class behavior images, this enforcement in KTH human body behavior database
The inventive method is tested respectively by example with the method that prior art uses, wherein Schindler, Ahmad, Jhuang and
Rodriguez method uses Split Method, the inventive method and Mikolajczyk method to use leaving-one method.The test knot of each method
As shown in table 1, the method for the present invention reaches 93.3% for the average recognition rate of each behavior to fruit, has exceeded the identification of additive method
Rate, has higher discrimination.
Each method discrimination in table 1:KTH data base
| Method | Evaluation of programme | Discrimination (%) |
| The inventive method | Leaving-one method | 93.3 |
| Schindler | Split Method | 90.73 |
| Ahmad | Split Method | 87.63 |
| Jhuang | Split Method | 91.68 |
| Rodriguez | Split Method | 88.66 |
| Mikolajczyk | Leaving-one method | 93.17 |
In sum, the method have the advantages that and calculate objective contour in consecutive frame by Euclidean distance
Change obtains profile energy variation rectangular histogram, and the profile energy variation utilizing k-means clustering method to obtain each two field picture is straight
Side's figure carries out unsupervised segmentation, can be improved the accuracy rate of identification to the method for classification results imparting weights by Geordie impurity level
And robustness, ensure that real-time simultaneously, solve tradition low based on model generalization ability in model matching method, antinoise
Ability and based on the problem of action classification similarity in state-space method.
The above is only the preferred embodiment of the present invention, it is noted that for the ordinary skill people of the art
For Yuan, on the premise of without departing from the technology of the present invention principle, it is also possible to make some improvement and modification, these improve and modification
Also should be regarded as protection scope of the present invention.
Claims (6)
1. the human motion recognition method in video image, is characterized in that, comprises the following steps:
Step one, carries out pretreatment to each frame input picture and obtains foreground area, foreground area is carried out screening and obtains target area
Territory；
Step 2, obtains objective contour according to target area；
Step 3, it is thus achieved that the profile energy variation rectangular histogram of X and Y-direction；
Step 4, is normalized profile energy variation rectangular histogram；
Step 5, the training stage: the training set forming profile energy variation rectangular histogram carries out the classification of motion, obtains human body behavior
Model also gives weights；
Step 6, at cognitive phase: the human body behavior model obtained profile energy histogram and the training stage of frame to be measured
Join, execution identification.
Human motion recognition method in a kind of video image the most according to claim 1, is characterized in that, in described step
In one, pretreatment uses background subtraction method, screening to use minimum enclosed rectangle frame method.
Human motion recognition method in a kind of video image the most according to claim 1, is characterized in that, in described step
In three, obtaining profile energy variation histogram method is:
31) the edge image I of adjacent two two field pictures is obtainededgeAnd Ilast_edge, use 10 × 10 windows by row traversal edge image
Iedge；
32) during traversal, when there is edge pixel in window, at previous frame image Ilast_edgeIn same area find therewith
The edge pixel of Euclidean distance minimum matches, and the value size of Euclidean distance changed as this point edge pixel energy；
33) after having traveled through, will row number as histogrammic abscissa, energy change value corresponding to each column is as histogrammic vertical
Coordinate, obtains profile energy variation rectangular histogram.
Human motion recognition method in a kind of video image the most according to claim 3, is characterized in that, in described step
In four, normalized process is, is first normalized rectangular histogram vertical coordinate so that it is value is between 0 to 1, then
It is the rectangular histogram of fixed size by Histogram Mapping a to abscissa.
Human motion recognition method in a kind of video image the most according to claim 1, is characterized in that, in described step
In five, sorting technique is:
51) utilizing k-means clustering method to obtain cluster barycenter collection, behavior carries out big class division, obtain respectively divides classification
Ci, wherein, 1≤i≤n, n are behavior classification numbers；
52) utilize Euclidean distance to each CiContrast two-by-two, obtain Geordie impurity level Gi, Geordie impurity level GiAs classification Ci's
Weights.
Human motion recognition method in a kind of video image the most according to claim 5, is characterized in that, in described step
In six, the detailed process in identification of behavior to be measured is:
61) for behavior S to be measuredq={ K1,K2,K3,.......,KlCarry out step one to three process and obtain its profile energy quantitative change
Change rectangular histogram, it is judged that image KtThe Euclidean distance of barycenter of rectangular histogram and each classification, choose the classification that Euclidean distance is minimum
As image KtAffiliated classification Ci, wherein, 1≤t≤l；
62) by SqThe probability belonging to each swooping template action behavior is set to Aq={ A1,A2,A3,......,An, wherein probability AiIt is
According to Geordie impurity level GiTo CiIt is optimized and obtains, Ai=Gi/Ci；
63) according to each two field picture generic Ai, select maximum AmaxSo that it is determined that SqType of action.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610621491.0A CN106295532B (en) | 2016-08-01 | 2016-08-01 | A kind of human motion recognition method in video image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610621491.0A CN106295532B (en) | 2016-08-01 | 2016-08-01 | A kind of human motion recognition method in video image |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106295532A true CN106295532A (en) | 2017-01-04 |
| CN106295532B CN106295532B (en) | 2019-09-24 |
Family
ID=57664039
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610621491.0A Expired - Fee Related CN106295532B (en) | 2016-08-01 | 2016-08-01 | A kind of human motion recognition method in video image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106295532B (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107272468A (en) * | 2017-08-01 | 2017-10-20 | 刘太龙 | Electronic security(ELSEC) based on communication ensures platform |
| CN107392945A (en) * | 2017-06-11 | 2017-11-24 | 杭州巨实科技有限公司 | A kind of two-dimensional silhouette matching process |
| CN108182416A (en) * | 2017-12-30 | 2018-06-19 | 广州海昇计算机科技有限公司 | A kind of Human bodys' response method, system and device under monitoring unmanned scene |
| CN109867186A (en) * | 2019-03-18 | 2019-06-11 | 浙江新再灵科技股份有限公司 | A kind of elevator malfunction detection method and system based on intelligent video analysis technology |
| CN110597251A (en) * | 2019-09-03 | 2019-12-20 | 三星电子（中国）研发中心 | Method and device for controlling intelligent mobile equipment |
| CN110866435A (en) * | 2019-08-13 | 2020-03-06 | 广州三木智能科技有限公司 | Far infrared pedestrian training method with self-similarity gradient oriented histogram |
| CN112070016A (en) * | 2020-09-08 | 2020-12-11 | 安徽兰臣信息科技有限公司 | Detection method for identifying child behavior and action |
| CN113221880A (en) * | 2021-04-29 | 2021-08-06 | 上海勃池信息技术有限公司 | OCR layout analysis method based on kini purity |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101576953A (en) * | 2009-06-10 | 2009-11-11 | 北京中星微电子有限公司 | Classification method and device of human body posture |
| CN101882217A (en) * | 2010-02-26 | 2010-11-10 | 杭州海康威视软件有限公司 | Target classification method of video image and device |
| CN102136066A (en) * | 2011-04-29 | 2011-07-27 | 电子科技大学 | Method for recognizing human motion in video sequence |
| CN102682302A (en) * | 2012-03-12 | 2012-09-19 | 浙江工业大学 | Human body posture identification method based on multi-characteristic fusion of key frame |
| CN103310233A (en) * | 2013-06-28 | 2013-09-18 | 青岛科技大学 | Similarity mining method of similar behaviors between multiple views and behavior recognition method |
| CN103400391A (en) * | 2013-08-09 | 2013-11-20 | 北京博思廷科技有限公司 | Multiple-target tracking method and device based on improved random forest |
| CN105139417A (en) * | 2015-07-27 | 2015-12-09 | 河海大学 | Method for real-time multi-target tracking under video surveillance |
-
2016
- 2016-08-01 CN CN201610621491.0A patent/CN106295532B/en not_active Expired - Fee Related
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101576953A (en) * | 2009-06-10 | 2009-11-11 | 北京中星微电子有限公司 | Classification method and device of human body posture |
| CN101882217A (en) * | 2010-02-26 | 2010-11-10 | 杭州海康威视软件有限公司 | Target classification method of video image and device |
| CN102136066A (en) * | 2011-04-29 | 2011-07-27 | 电子科技大学 | Method for recognizing human motion in video sequence |
| CN102682302A (en) * | 2012-03-12 | 2012-09-19 | 浙江工业大学 | Human body posture identification method based on multi-characteristic fusion of key frame |
| CN103310233A (en) * | 2013-06-28 | 2013-09-18 | 青岛科技大学 | Similarity mining method of similar behaviors between multiple views and behavior recognition method |
| CN103400391A (en) * | 2013-08-09 | 2013-11-20 | 北京博思廷科技有限公司 | Multiple-target tracking method and device based on improved random forest |
| CN105139417A (en) * | 2015-07-27 | 2015-12-09 | 河海大学 | Method for real-time multi-target tracking under video surveillance |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107392945A (en) * | 2017-06-11 | 2017-11-24 | 杭州巨实科技有限公司 | A kind of two-dimensional silhouette matching process |
| CN107392945B (en) * | 2017-06-11 | 2019-10-25 | 杭州巨梦科技有限公司 | A kind of two-dimensional silhouette matching process |
| CN107272468A (en) * | 2017-08-01 | 2017-10-20 | 刘太龙 | Electronic security(ELSEC) based on communication ensures platform |
| CN108182416A (en) * | 2017-12-30 | 2018-06-19 | 广州海昇计算机科技有限公司 | A kind of Human bodys' response method, system and device under monitoring unmanned scene |
| CN109867186A (en) * | 2019-03-18 | 2019-06-11 | 浙江新再灵科技股份有限公司 | A kind of elevator malfunction detection method and system based on intelligent video analysis technology |
| CN109867186B (en) * | 2019-03-18 | 2020-11-10 | 浙江新再灵科技股份有限公司 | Elevator trapping detection method and system based on intelligent video analysis technology |
| CN110866435A (en) * | 2019-08-13 | 2020-03-06 | 广州三木智能科技有限公司 | Far infrared pedestrian training method with self-similarity gradient oriented histogram |
| CN110866435B (en) * | 2019-08-13 | 2023-09-12 | 广州三木智能科技有限公司 | Far infrared pedestrian training method for self-similarity gradient orientation histogram |
| CN110597251A (en) * | 2019-09-03 | 2019-12-20 | 三星电子（中国）研发中心 | Method and device for controlling intelligent mobile equipment |
| CN110597251B (en) * | 2019-09-03 | 2022-10-25 | 三星电子（中国）研发中心 | Method and device for controlling intelligent mobile equipment |
| CN112070016A (en) * | 2020-09-08 | 2020-12-11 | 安徽兰臣信息科技有限公司 | Detection method for identifying child behavior and action |
| CN112070016B (en) * | 2020-09-08 | 2023-12-26 | 浙江铂视科技有限公司 | Detection method for identifying child behavior and action |
| CN113221880A (en) * | 2021-04-29 | 2021-08-06 | 上海勃池信息技术有限公司 | OCR layout analysis method based on kini purity |
| CN113221880B (en) * | 2021-04-29 | 2022-08-05 | 上海勃池信息技术有限公司 | OCR layout analysis method based on kini purity |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106295532B (en) | 2019-09-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106295532B (en) | A kind of human motion recognition method in video image | |
| CN108921083B (en) | Illegal mobile vendor identification method based on deep learning target detection | |
| CN102609686B (en) | Pedestrian detection method | |
| CN106529499A (en) | Fourier descriptor and gait energy image fusion feature-based gait identification method | |
| CN105160310A (en) | 3D (three-dimensional) convolutional neural network based human body behavior recognition method | |
| CN102214309B (en) | Special human body recognition method based on head and shoulder model | |
| CN102609720B (en) | Pedestrian detection method based on position correction model | |
| CN107239762A (en) | Patronage statistical method in a kind of bus of view-based access control model | |
| CN103226835B (en) | Based on method for tracking target and the system of online initialization gradient enhancement regression tree | |
| CN108549864A (en) | Area-of-interest filter method towards vehicle-mounted thermal imaging pedestrian detection and device | |
| CN105512684A (en) | Vehicle logo automatic identification method based on principal component analysis convolutional neural network | |
| CN110119726A (en) | A kind of vehicle brand multi-angle recognition methods based on YOLOv3 model | |
| CN107491749B (en) | Method for detecting global and local abnormal behaviors in crowd scene | |
| CN105956560A (en) | Vehicle model identification method based on pooling multi-scale depth convolution characteristics | |
| CN104281853A (en) | Behavior identification method based on 3D convolution neural network | |
| CN103942577A (en) | Identity identification method based on self-established sample library and composite characters in video monitoring | |
| CN104751136A (en) | Face recognition based multi-camera video event retrospective trace method | |
| CN104778474A (en) | Classifier construction method for target detection and target detection method | |
| CN104978567A (en) | Vehicle detection method based on scenario classification | |
| CN110991397B (en) | Travel direction determining method and related equipment | |
| CN107808376A (en) | A kind of detection method of raising one's hand based on deep learning | |
| CN102663411A (en) | Recognition method for target human body | |
| CN104966305A (en) | Foreground detection method based on motion vector division | |
| CN104537356A (en) | Pedestrian re-identification method and device for carrying out gait recognition through integral scheduling | |
| CN113205107A (en) | Vehicle type recognition method based on improved high-efficiency network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20190924 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |