CN100568266C - A kind of abnormal behaviour detection method based on the sports ground partial statistics characteristic analysis - Google Patents
A kind of abnormal behaviour detection method based on the sports ground partial statistics characteristic analysis Download PDFInfo
- Publication number
- CN100568266C CN100568266C CNB2008101008611A CN200810100861A CN100568266C CN 100568266 C CN100568266 C CN 100568266C CN B2008101008611 A CNB2008101008611 A CN B2008101008611A CN 200810100861 A CN200810100861 A CN 200810100861A CN 100568266 C CN100568266 C CN 100568266C
- Authority
- CN
- China
- Prior art keywords
- motion
- feature
- analysis
- image
- sports ground
- 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.)
- Expired - Fee Related
Links
Images
Landscapes
- Image Analysis (AREA)
Abstract
The present invention relates to a kind of abnormal behaviour detection method based on the sports ground partial statistics characteristic analysis.This system mainly comprises the sports ground analysis of video image, and partial statistics characteristic extracts, based on the statistical learning and the mode identification technology of sample.At first, extract the feature of object in the image, calculate its motion state, form a sports ground by basic motion analysis technique.On this basis, local movable information is carried out statistical nature extract, obtain the local motion feature of sports ground.Space distribution relation table with these motion features is shown as global structure information at last, adopts the method based on statistical learning, discerns the type of behavior.This algorithm improves the efficient and the robustness of algorithm by directly carrying out behavior identification based on the analysis of movable information.
Description
Technical field
The present invention relates to computer monitoring technology, based on the mode identification technology of statistical learning with based on the signature analysis technology of local motion field based on moving image.Be a kind of vision monitoring content analysis method, belong to computer vision, Intelligent Information Processing field.
Background technology
Along with the attention of society to the public safety problem, monitoring has in real time obtained application more and more widely.At present the problem of monitoring is that a large amount of monitor messages is difficult to be handled timely and effectively, utilizes the computer vision technique analysis and understands people's motion, and record is provided and reports to the police, and then helps to improve the security monitoring level of public place.Assist identification by computing machine, become a hot issue of computer vision field human behavior and incident.
The Intellectual Analysis Technology of vision monitoring is the focus and the difficult point problem of computer vision field, relates to advanced subject such as Flame Image Process, machine learning.The correlative study of academicly carrying out in recent years is a lot, comprises the research of the intelligent monitoring project of country 863,973.Whether the abnormal behaviour detection is meant is at first analyzed respectively and modeling self-defining regular event behavior and other behavioral datas, come the differentiation behavior to have unusually according to behavioral test to both similar programs then.Present research mainly concentrates moving region, speed by human body whether to satisfy restrictions such as predetermined condition, detects elementary events such as the personage occurs, crosses the border.Company such as U.S. ObjectVideo has developed corresponding product on this basis, and has obtained certain application.
Yet also there are certain problem in these researchs and application: at first, present research is based on the detection and tracking information of foreground area more, and then analyze and whether satisfy predetermined condition, make problem be confined to the detection and tracking of object, and shorter mention has certain limitation to the understanding of object of which movement action and rule thereof on the function.Secondly, part vision understanding method adopts the manikin method of machinery, by the match video image, recovers the state of human synovial motion, and then understands the motion of human body.Handling has like this increased intermediate link, makes model complicated more and be difficult to accurate realization.
Summary of the invention
The present invention is by the analysis to the human motion field, extract local motion feature, and the feature and the distribution thereof of comprehensive zones of different, obtain the description of individual movement, at last a large amount of normal and abnormal motion data are carried out machine learning, set up the human motion characteristic model, realize the real-time detection of human body abnormal behaviour.
The present invention relates generally to computer vision and area of pattern recognition, obtain the essential information of human motion field by video image, on this basis, at first at the characteristics of motion complexity, the human motion of complexity is decomposed into simple local motion, by location and analysis local motion field, extract local motion feature; The motion feature and the distribution characteristics thereof of comprehensive human body zones of different obtain the description of individual movement then; At last a large amount of normal and abnormal motion data are carried out machine learning, set up the human motion characteristic model, realize the real-time detection of human body abnormal behaviour.
The entire method process is as follows: at first by some groups of the predefined dissimilar actions of video acquisition, as training sample; According to the human motion area detection result, regional area is divided then; Set up the optical flow field of motion,, obtain the description of local motion by location, local motion field and analytical approach; Extract provincial characteristics, and incorporate the right space distribution relation of feature, be expressed as global feature; To these features, carry out a large amount of study then, set up the SVR model; The video data of test phase by gathering in real time carries out the feature extraction of above step to uncertain action, judges the probability that belongs to a certain class action by Bayesian method, and the type of action of choosing the probability maximum at last is as recognition result.
(1) based on the signature analysis of local motion field.
At first, provide the basic light stream data of sports ground by the analysis of video image, and given image I (x, y, t), expression any time t, image coordinate (x, the brightness value I that y) locates.Then according to basic light stream formula (1), on the indeclinable basis of the characteristics of luminescence of supposition human body, we can obtain, and (x y) locates the movement velocity V of unique point.
U wherein, v is the image sampling coordinate axis.
The motion feature L={V of light stream
1, V
2Obtain by the optical flow field vector being carried out Gauss's cluster, be expressed as two main self-movement component V
1, V
2, to simplify the expression of whole optical flow field.
The space distribution of light stream realizes by the calculating of square, can obtain basic moment characteristics according to formula (2), wherein zeroth order moment characteristics M
0,0The area of expression moving region, first moment feature M
0,1And M
1,0Represent the gravity centre distribution situation of two change in coordinate axis direction respectively, more the high-order feature is then represented the distributed intelligence of higher frequency scope, so just can represent the distribution characteristics J={M of moving region by a series of squares and conversion thereof
00, M
01, M
10....
M
pq＝∑
x∑
y||V||(x，y)x
py
q (2)
Because the complicacy of human motion, motion of overall importance and distributional analysis thereof can only obtain overall movement tendency, and its effect is very limited; Because the design feature of human body has the continuity of tangible overall uncertainty and local regional movement, we localize to above optical flow analysis and moving region analysis, promptly carry out feature calculation in the given area in addition.
(2) overall human motion character description method
The motion feature of the overall situation the more important thing is by local feature and distributes and describe global feature that concrete grammar is as follows except the description of local feature:
At first analyze the distribution of moving region by background and present frame difference image, wherein the differences in motion partial image can directly obtain according to formula (3).
D(x，y，t)＝I(x，y，t)-I(x，y，t
0) (3)
I (x, y, t wherein
0) be background image; Calculate the border of this moving region then,,, the human motion zone is divided into independently r part in advance, represent the motion conditions of the local limbs of people region respectively in conjunction with the design feature of human body to determine the position at human body place; In each zone, all need independent calculating optical flow field like this, and the light stream motion feature L in the regional i
iWith distribution of movement feature J
i, i ∈ N wherein, i≤r;
In order to describe the relative position in different characteristic zone, we have defined feature to T
q=＜J
1, J
q, representation feature J
1With J
qThe formed vector of center line, final human motion global characteristics may be encoded as:
g＝[L
1，…，L
r，J
1，…，J
r，T
2，…，T
r]。
(3) based on Bayesian action sequence identification
At first define the object set H={h of action sequence identification
0, h
1H
n, for each behavior aggregate G={G
0, G
1G
nIn each action all gather a certain amount of sample, set up corresponding SVR (support vector regression) model.A so given new action sequence g
i, at first can obtain to belong to the instant probability P (g of certain target by the SVR model
i| h
m), by all targets being subordinate to the comparison of probability, the instant Motion Recognition result who is model of probability maximum.
Because instant result can not react whole motion process,, need to calculate an action sequence g in order to reduce error rate
t, g
(t-1)... belong to any target h
lProbability P (h
l| g
t, g
(t-1)...).
According to the continuity of state, we can regard motion change as a Markovian process, think that its state is only relevant with current action with last state, then
P(h
l|g
t，g
(t-1)，......)＝P(h
l|h
(l-1)，g
t)
Suppose that in addition it is independently then that t-1 state and t move constantly
P(h
l|h
(l-1)，g
t)＝P(h
l|h
(l-1))P(h
l|g
t)
According to Bayesian formula:
P(h
l|g
t)＝P(h
l)P(g
t|h
l)/∑
kP(g
t|h
k)P(h
k)
Given state transition matrix P (h
l| h
(l-1)) and prior probability P (h
l), just can obtain the recognition result of whole sequence according to the SVR model.Wherein the frequency statistics that occurs according to exercises in the training sample of state-transition matrix and prior probability obtains, and also can adopt equally distributed strategy to carry out.
Description of drawings:
Fig. 1 is a system of the present invention operational scheme.
Fig. 2 is an algorithm flow of the present invention.
Fig. 3 is that sports ground of the present invention is analyzed synoptic diagram, and wherein square frame is the regional area of being divided, and red circle is represented detected local motion feature center, and red line is represented its speed, and the blue line between circle is represented the proper vector of global characteristics.
Embodiment:
Below in conjunction with accompanying drawing embodiments of the present invention are described in detail.
Fig. 1 is system's operational flow diagram, and Fig. 2 is the process flow diagram of feature extraction algorithm, and the inventive method is according to Fig. 1 flow process, and system's operation comprises following concrete steps:
1. sample video data acquiring: native system adopts the machine learning principle to carry out the modeling and the identification of human motion, therefore needs work such as study that a large amount of motion samples are correlated with and verification.We are example with 5 kinds of specific actions, and each has gathered the plurality of sections video sequence, and get wherein a part and learn as training set, and a part is carried out model checking as test set, has made up samples of video data.
2. graphical analysis and feature extraction: concrete leaching process flow process is as shown in Figure 2 carried out, and is described below.
1) obtains video data: comprise the video of gathering in real time in training video and the detection.
2) sports ground calculates: (x, y t), according to basic light stream formula (1), calculate the movement velocity of all unique points to given video image I, wherein comprise size and Orientation information.
3) human detection and area dividing: at first (x, y is t) with background image I (x, y, t by present image I
0) method of difference, obtain the zone of human motion.Calculate this regional center and square boundary information, quad-tree partition is carried out in the region, four local and initial zones that obtain moving.
4) local feature extracts: local feature extracts and comprises two aspect features, and one is the principal component analysis (PCA) L of vector, to the velocity of all unique points, by the method for cluster, obtains two main speed components; To calculating, obtain to comprise in the characteristic area moment characteristics J of motion center of gravity on the other hand by regional area internal moment to velocity magnitude.
5) global characteristics analysis: adopt to be described based on principal and subordinate's feature mode method, at first, we select a main feature, mate with other features respectively, form 3 features to T, represent with the form of vector, comprise distance and angle information between feature.Structural information between feature and local feature are carried out sequential encoding promptly obtain the global characteristics g that moves.
6) model learning and identification: all at last training are all carried out based on such global motion feature with study.
3. behavior modeling: at first all gather a certain amount of sample,, obtain the human motion global characteristics, set up corresponding SVR model by graphical analysis and feature extraction for each action.Pass through the change calculations state-transition matrix P (h of status switch simultaneously
l| h
(l-1)) and prior probability P (h
l).
4. set up model bank: so far finish whole off-line training process, SVR model and state-transition matrix P (h that training is obtained
l| h
(l-1)) and prior probability P (h
l) be saved in the database.
5. real-time data acquisition: in application process, at first want above SVR model of initialization and correlation parameter, gather motion video sequence analysis in real time by camera.
6. adopt the method identical to carry out feature extraction with 2.
7. behavior identification: a given new action sequence g
t, g
T-1..., need ask to belong to any one dbjective state h
lProbability P (h
l| g
t, g
T-1...).We regard it as a Markovian process, think that its state is only relevant with current action with last state, then by Bayesian formula:
P(h
l|g
t)＝P(h
l)P(g
t|h
l)/∑
kP(g
t|h
k)P(h
k)
Given state transition matrix P (h
l| h
(l-1)) and prior probability P (h
l), just can obtain the similarity to a certain action of whole sequence according to the SVR model.
8. interpretation of result: compare by similarity at last each behavior, choose satisfy maximal phase that threshold value (Y=0.5) requires like behavior as recognition result, otherwise think and do not belong to above training action, promptly be judged to be abnormal behaviour.
Claims (1)
1. the abnormal behaviour detection method based on the sports ground partial statistics characteristic analysis is characterized in that, said method comprising the steps of:
(1) sample video data acquiring: by some groups of the predefined dissimilar actions of video acquisition, as training sample;
(2) graphical analysis and feature extraction may further comprise the steps:
I) obtain video data: comprise the video of gathering in real time in training video and the detection;
Ii) sports ground calculates: at first by the analysis of video image, provide the basic light stream data of sports ground, given image I (x, y, t), expression any time t, image coordinate (x, the brightness value I that y) locates; Then according to basic light stream formula (1), on the indeclinable basis of the characteristics of luminescence of supposition human body, we can obtain (x, y) locate the movement velocity V of unique point:
U wherein, v is the image sampling coordinate axis;
Iii) human detection and area dividing: at first analyze the distribution of moving region by background and present frame difference image, wherein the differences in motion partial image can directly obtain according to formula (2):
D(x，y，t)＝I(x，y，t)-I(x，y，t
0) (2)
I (x, y, t wherein
0) be background image; Calculate the border of this moving region then,,, the human motion zone is divided into independently r part in advance, represent the motion conditions of the local limbs of people region respectively in conjunction with the design feature of human body to determine the position at human body place; In each zone, all need independent calculating optical flow field like this, and the light stream motion feature L in the regional i
iWith distribution of movement feature J
i, i ∈ N wherein, i≤r;
Iv) local feature extracts: the motion feature L={V of light stream
1, V
2Obtain by the optical flow field vector being carried out Gauss's cluster, be expressed as two main self-movement component V
1, V
2, to simplify the expression of whole optical flow field; The space distribution of light stream realizes by the calculating of square, can obtain basic moment characteristics according to formula (3), wherein zeroth order moment characteristics M
0,0The area of expression moving region, first moment feature M
0,1And M
1,0Represent the gravity centre distribution situation of two change in coordinate axis direction respectively, more the high-order feature is then represented the distributed intelligence of higher frequency scope, so just can represent the distribution characteristics J={M of moving region by a series of squares and conversion thereof
00, M
01, M
10...;
M
pq＝∑
x∑
y||V||(x，y)x
py
q (3)
V) global characteristics analysis: adopt to be described based on principal and subordinate's feature mode method, in order to describe the relative position in different characteristic zone, defined feature is to T
q=＜J
1, J
q, representation feature J
1With J
qThe formed vector of center line, final human motion global characteristics may be encoded as: g=[L
1..., L
r, J
1..., J
r, T
2..., T
r];
Vi) model learning and identification: all at last training are all carried out based on such global motion feature with study;
(3) behavior modeling may further comprise the steps:
At first define the object set H={h of action sequence identification
0, h
1H
n, for each behavior aggregate G={G
0, G
1G
nIn each action all gather a certain amount of sample, set up corresponding support vector regression model; A so given new action sequence g
t, at first can obtain to belong to the instant probability P (g of certain target by the support vector regression model
t| h
m), by all targets being subordinate to the comparison of probability, the instant Motion Recognition result who is model of probability maximum;
Because instant result can not react whole motion process,, need to calculate an action sequence g in order to reduce error rate
t, g
(t-1)... belong to any target h
1Probability P (h
l| g
t, g
(t-1)...);
According to the continuity of state, we can regard motion change as a Markovian process, think that its state is only relevant with current action with last state, then
P(h
l|g
t，g
(t-1)，……)＝P(h
l|h
(l-1)，g
t)
Suppose that in addition it is independently then that t-1 state and t move constantly
P(h
l|h
(l-1)，g
t)＝P(h
l|h
(l-1))P(h
l|g
t)
According to Bayesian formula:
P(h
l|g
t)＝P(h
l)P(g
t|h
l)/∑
kP(g
t|h
k)P(h
k)
Given state transition matrix P (h
l| h
(l-1)) and prior probability P (h
l), just can obtain the recognition result of whole sequence according to the support vector regression model;
(4) set up model bank: support vector regression model and state-transition matrix P (h that training is obtained
l| h
(l-1)) and prior probability P (h
l) be saved in the database;
(5) real-time data acquisition: gather motion video sequence analysis in real time by camera;
(6) adopt the method identical to carry out feature extraction with step (2);
(7) behavior identification: a given new action sequence, by Bayesian formula:
P(h
l|g
t)＝P(h
l)P(g
t|h
l)/∑
kP(g
t|h
k)P(h
k)
Given state transition matrix P (h
l| h
(l-1)) and prior probability P (h
l), just can obtain the similarity to a certain action of whole sequence according to the support vector regression model;
(8) interpretation of result: compare by similarity at last each behavior, choose satisfy maximal phase that threshold value requires like behavior as recognition result, otherwise think and do not belong to above training action, promptly be judged as abnormal behaviour.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2008101008611A CN100568266C (en) | 2008-02-25 | 2008-02-25 | A kind of abnormal behaviour detection method based on the sports ground partial statistics characteristic analysis |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2008101008611A CN100568266C (en) | 2008-02-25 | 2008-02-25 | A kind of abnormal behaviour detection method based on the sports ground partial statistics characteristic analysis |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN101271527A CN101271527A (en) | 2008-09-24 |
| CN100568266C true CN100568266C (en) | 2009-12-09 |
Family
ID=40005488
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNB2008101008611A Expired - Fee Related CN100568266C (en) | 2008-02-25 | 2008-02-25 | A kind of abnormal behaviour detection method based on the sports ground partial statistics characteristic analysis |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN100568266C (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107103614A (en) * | 2017-04-12 | 2017-08-29 | 合肥工业大学 | The dyskinesia detection method encoded based on level independent element |
Families Citing this family (54)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP5421072B2 (en) * | 2008-11-19 | 2014-02-19 | クラリオン株式会社 | Approaching object detection system |
| US8165349B2 (en) * | 2008-11-29 | 2012-04-24 | International Business Machines Corporation | Analyzing repetitive sequential events |
| CN101739685B (en) * | 2009-02-11 | 2012-04-18 | 北京智安邦科技有限公司 | Moving object classification method and system thereof |
| EP2415022A4 (en) | 2009-04-01 | 2016-12-14 | I-Cetana Pty Ltd | Systems and methods for detecting anomalies from data |
| CN101714264B (en) * | 2009-11-20 | 2011-08-24 | 西安电子科技大学 | Passive detection method and system of scales jump actions in highway charging and weighing |
| CN101719216B (en) * | 2009-12-21 | 2012-01-04 | 西安电子科技大学 | Movement human abnormal behavior identification method based on template matching |
| CN101872418B (en) * | 2010-05-28 | 2012-09-12 | 电子科技大学 | Detection method based on group environment abnormal behavior |
| US20130130214A1 (en) * | 2010-08-04 | 2013-05-23 | Takeo Ohno | Behavior characteristic extraction device, a behavior characteristic extraction system, a behavior characteristic extraction method and a behavior characteristic extraction program |
| CN101968848B (en) * | 2010-09-27 | 2013-01-16 | 哈尔滨工业大学深圳研究生院 | Video monitoring method and system and video monitoring alarm system |
| CN102254329A (en) * | 2011-08-18 | 2011-11-23 | 上海方奥通信技术有限公司 | Abnormal behavior detection method based on motion vector classification analysis |
| CN103150718B (en) * | 2011-12-06 | 2016-02-10 | 中国科学院电子学研究所 | Based on the method for detecting change of remote sensing image of Change vector Analysis and classification and predicting |
| CN102789672B (en) * | 2012-07-03 | 2017-10-10 | 深圳市瑞工科技有限公司 | A kind of baby's sleeping position intelligent identification Method and device |
| CN103544466A (en) * | 2012-07-09 | 2014-01-29 | 西安秦码软件科技有限公司 | Vector field model based behavior analysis method |
| CN103810717B (en) * | 2012-11-09 | 2016-08-24 | 浙江大华技术股份有限公司 | A kind of human body behavioral value method and device |
| CN102938078A (en) * | 2012-11-30 | 2013-02-20 | 北京理工大学 | Unbalanced timing sequence-based monitoring video abnormal event detection method |
| CN104052617A (en) * | 2013-03-13 | 2014-09-17 | 北京千橡网景科技发展有限公司 | Method and device for user behavior analysis based on continuous-time Markov chain |
| CN103400160B (en) * | 2013-08-20 | 2017-03-01 | 中国科学院自动化研究所 | A kind of zero training sample Activity recognition method |
| CN103473533B (en) * | 2013-09-10 | 2017-03-15 | 上海大学 | Moving Objects in Video Sequences abnormal behaviour automatic testing method |
| US11861906B2 (en) | 2014-02-28 | 2024-01-02 | Genius Sports Ss, Llc | Data processing systems and methods for enhanced augmentation of interactive video content |
| US10713494B2 (en) | 2014-02-28 | 2020-07-14 | Second Spectrum, Inc. | Data processing systems and methods for generating and interactive user interfaces and interactive game systems based on spatiotemporal analysis of video content |
| US11120271B2 (en) | 2014-02-28 | 2021-09-14 | Second Spectrum, Inc. | Data processing systems and methods for enhanced augmentation of interactive video content |
| EP3111659A4 (en) * | 2014-02-28 | 2017-12-13 | Second Spectrum Inc. | System and method for performing spatio-temporal analysis of sporting events |
| US10521671B2 (en) | 2014-02-28 | 2019-12-31 | Second Spectrum, Inc. | Methods and systems of spatiotemporal pattern recognition for video content development |
| US10769446B2 (en) | 2014-02-28 | 2020-09-08 | Second Spectrum, Inc. | Methods and systems of combining video content with one or more augmentations |
| CN103854016B (en) * | 2014-03-27 | 2017-03-01 | 北京大学深圳研究生院 | Jointly there is human body behavior classifying identification method and the system of feature based on directivity |
| CN104156979B (en) * | 2014-07-25 | 2016-09-14 | 南京大学 | Deviant Behavior online test method in a kind of video based on gauss hybrid models |
| CN104299355A (en) * | 2014-09-28 | 2015-01-21 | 厦门蓝斯通信股份有限公司 | Vehicle monitoring method and system based on video intelligent recognition |
| CN104615711A (en) * | 2015-02-04 | 2015-05-13 | 中国科学院自动化研究所 | Human movement labeling method based on Bayesian network |
| CN105046285B (en) * | 2015-08-31 | 2018-08-17 | 武汉鹰视智能科技有限公司 | A kind of abnormal behaviour discrimination method based on kinematic constraint |
| CN105184257B (en) * | 2015-09-08 | 2018-08-07 | 北京航空航天大学 | Object detection method and device |
| CN105718857B (en) * | 2016-01-13 | 2019-06-04 | 兴唐通信科技有限公司 | A kind of human body anomaly detection method and system |
| CN105825512B (en) * | 2016-03-18 | 2018-06-29 | 武汉大学 | The abnormal target in hyperspectral remotely sensed image object detection method returned based on steady background |
| CN105912652B (en) * | 2016-04-08 | 2019-05-31 | 华南师范大学 | Anomaly detection method and system based on correlation rule and user property |
| WO2017206005A1 (en) * | 2016-05-30 | 2017-12-07 | 中国石油大学(华东) | System for recognizing postures of multiple people employing optical flow detection and body part model |
| CN106060653A (en) * | 2016-06-08 | 2016-10-26 | 武汉鸿瑞达信息技术有限公司 | Method for filtering improper action in live internet video |
| CN106407889B (en) * | 2016-08-26 | 2020-08-04 | 上海交通大学 | Method for recognizing human body interaction in video based on optical flow graph deep learning model |
| CN107392100B (en) * | 2017-06-17 | 2020-07-07 | 复旦大学 | Detection method for automatically detecting local abnormality in monitoring video |
| CN107943077A (en) * | 2017-11-24 | 2018-04-20 | 歌尔股份有限公司 | A kind of method for tracing, device and the unmanned plane of unmanned plane drop target |
| CN108151732A (en) * | 2017-12-22 | 2018-06-12 | 浙江西湖高等研究院 | A kind of long-range position and behavior method of estimation |
| CN108226915B (en) * | 2017-12-25 | 2021-07-30 | 中国人民解放军63921部队 | Quantitative representation space multi-target spatial distribution method |
| CN108211318B (en) * | 2018-01-23 | 2019-08-23 | 北京易智能科技有限公司 | Based on the race walking posture analysis method perceived in many ways |
| CN108596019B (en) * | 2018-03-12 | 2022-01-28 | 杭州英歌智达科技有限公司 | Camera monitoring method and system based on diagonal potential |
| CN110580708B (en) * | 2018-06-11 | 2022-05-31 | 杭州海康威视数字技术股份有限公司 | Rapid movement detection method and device and electronic equipment |
| CN108985183A (en) * | 2018-06-26 | 2018-12-11 | 深圳市唯特视科技有限公司 | A kind of behavioral value framework based on real-time end-to-end binary-flow network |
| CN109727270B (en) * | 2018-12-10 | 2021-03-26 | 杭州帝视科技有限公司 | Motion mechanism and texture feature analysis method and system of cardiac nuclear magnetic resonance image |
| CN109922310B (en) * | 2019-01-24 | 2020-11-17 | 北京明略软件系统有限公司 | Target object monitoring method, device and system |
| CN110786869B (en) * | 2019-10-29 | 2021-12-21 | 浙江工业大学 | Method for detecting fatigue degree of programmer |
| CN110852247A (en) * | 2019-11-07 | 2020-02-28 | 北京云迹科技有限公司 | Abnormality detection method, abnormality detection device, electronic apparatus, and computer-readable storage medium |
| US11113535B2 (en) | 2019-11-08 | 2021-09-07 | Second Spectrum, Inc. | Determining tactical relevance and similarity of video sequences |
| CN111062297B (en) * | 2019-12-11 | 2023-05-23 | 青岛科技大学 | Violent abnormal behavior detection method based on EANN deep learning model |
| CN110889395B (en) * | 2019-12-12 | 2023-08-15 | 广州中科永信科技有限公司 | Machine learning-based mechanical motion recognition method and system |
| CN111062356B (en) * | 2019-12-26 | 2024-03-26 | 沈阳理工大学 | Method for automatically identifying abnormal human body actions from monitoring video |
| CN116910824B (en) * | 2023-08-28 | 2024-02-06 | 广东中山网传媒信息科技有限公司 | Safety big data analysis method and system based on distributed multi-source measure |
| CN117132936A (en) * | 2023-08-31 | 2023-11-28 | 北京中电拓方科技股份有限公司 | Data carding and data access system of coal plate self-building system |
-
2008
- 2008-02-25 CN CNB2008101008611A patent/CN100568266C/en not_active Expired - Fee Related
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107103614A (en) * | 2017-04-12 | 2017-08-29 | 合肥工业大学 | The dyskinesia detection method encoded based on level independent element |
| CN107103614B (en) * | 2017-04-12 | 2019-10-08 | 合肥工业大学 | Dyskinesia detection method based on level independent element coding |
Also Published As
| Publication number | Publication date |
|---|---|
| CN101271527A (en) | 2008-09-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN100568266C (en) | A kind of abnormal behaviour detection method based on the sports ground partial statistics characteristic analysis | |
| Feng et al. | Spatio-temporal fall event detection in complex scenes using attention guided LSTM | |
| CN106875424B (en) | A kind of urban environment driving vehicle Activity recognition method based on machine vision | |
| CN106127802B (en) | A kind of movement objective orbit method for tracing | |
| CN102663409B (en) | Pedestrian tracking method based on HOG-LBP | |
| Rasouli | Deep learning for vision-based prediction: A survey | |
| CN102663429B (en) | Method for motion pattern classification and action recognition of moving target | |
| CN103077423B (en) | To run condition detection method based on crowd's quantity survey of video flowing, local crowd massing situation and crowd | |
| CN102682302A (en) | Human body posture identification method based on multi-characteristic fusion of key frame | |
| CN104680557A (en) | Intelligent detection method for abnormal behavior in video sequence image | |
| CN102592138B (en) | Object tracking method for intensive scene based on multi-module sparse projection | |
| CN104091348A (en) | Multi-target tracking method integrating obvious characteristics and block division templates | |
| CN101098465A (en) | Moving object detecting and tracing method in video monitor | |
| CN104200488A (en) | Multi-target tracking method based on graph representation and matching | |
| CN103605986A (en) | Human motion recognition method based on local features | |
| CN103426179A (en) | Target tracking method and system based on mean shift multi-feature fusion | |
| CN101799968A (en) | Detection method and device for oil well intrusion based on video image intelligent analysis | |
| CN104318588A (en) | Multi-video-camera target tracking method based on position perception and distinguish appearance model | |
| Cao et al. | NeuroGrasp: multimodal neural network with Euler region regression for neuromorphic vision-based grasp pose estimation | |
| CN104159088A (en) | System and method of remote monitoring of intelligent vehicle | |
| Chen et al. | Crowd escape behavior detection and localization based on divergent centers | |
| Zhao et al. | Detection of passenger flow on and off buses based on video images and YOLO algorithm | |
| CN104463909A (en) | Visual target tracking method based on credibility combination map model | |
| CN106447698A (en) | Multi-pedestrian tracking method and system based on distance sensor | |
| Batool et al. | Telemonitoring of daily activities based on multi-sensors data fusion |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| C17 | Cessation of patent right | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20091209 Termination date: 20110225 |