CN109086717A - Act of violence detection system and method based on human skeleton and motor message feature - Google Patents
Act of violence detection system and method based on human skeleton and motor message feature Download PDFInfo
- Publication number
- CN109086717A CN109086717A CN201810864496.5A CN201810864496A CN109086717A CN 109086717 A CN109086717 A CN 109086717A CN 201810864496 A CN201810864496 A CN 201810864496A CN 109086717 A CN109086717 A CN 109086717A
- Authority
- CN
- China
- Prior art keywords
- motor message
- act
- module
- feature
- human skeleton
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- 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
- G06V20/42—Higher-level, semantic clustering, classification or understanding of video scenes, e.g. detection, labelling or Markovian modelling of sport events or news items of sport video content
-
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- 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
- G06V40/20—Movements or behaviour, e.g. gesture recognition
- G06V40/23—Recognition of whole body movements, e.g. for sport training
-
- 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
- G06V20/44—Event detection
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Multimedia (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Psychiatry (AREA)
- Social Psychology (AREA)
- Human Computer Interaction (AREA)
- Probability & Statistics with Applications (AREA)
- Computational Linguistics (AREA)
- Software Systems (AREA)
- Image Analysis (AREA)
Abstract
The present invention relates to video image processing technology, the act of violence detection method based on human skeleton and motor message feature that it discloses a kind of improves the real-time and accuracy of detection, and adapts to a variety of different monitoring environment.This method comprises: a. is continuously extracted 3 frames by certain time interval, the motion vector of the key position pixel of human skeleton is then sought according to the frame difference of adjacent two frame, extracts the pixel wherein moved, and carry out moving region division；B. according in every figure moving region quantity and direction classify to type of sports；C. the statistic of the motor message of picture under each operation type is classified is calculated；D., the statistic of motor message is input to the prediction that act of violence is carried out in trained SVM model as feature.In addition, the present invention also provides a kind of act of violence detection system based on human skeleton and motor message feature, is suitable for plurality of application scenes.
Description
Technical field
The present invention relates to video image processing technologies, and in particular to a kind of sudden and violent based on human skeleton and motor message feature
Lixing is detection system and method.
Background technique
With a large amount of uses of monitoring system, there is fulminant growth in video data.The effect of monitoring system is to carry out
Target detection and unusual checking.With the sharp increase of data, the mode that traditional dependence manually monitors is more tired
Difficulty, and inefficiency.Therefore, become hot spot by the research of the monitoring system of artificial intelligence.Wherein, for the violence row of people
For detection be very important research direction.
Due to act of violence movement compared with simply run, slip a line for than it is complicated very much, so how to carry out act of violence
Detection is also the difficult point of correlative study.Currently, traditional act of violence detection mainly uses the side based on artificial design features
Method also has certain defects although recognition accuracy is higher, such as: real-time effect cannot be reached, the shadow vulnerable to noise
Ring etc..
Summary of the invention
The technical problems to be solved by the present invention are: proposing a kind of violence row based on human skeleton and motor message feature
For detection system and method, the real-time and accuracy of detection are improved, and adapts to a variety of different monitoring environment.
The technical proposal adopted by the invention to solve the above technical problems is that: based on human skeleton and motor message feature
Act of violence detection system, comprising: Acquiring motion area module, type of sports categorization module, motor message statistical module and figure
As prediction module；
The Acquiring motion area module, for continuously extracting 3 frames by certain time interval；Then according to adjacent two frame
Frame difference seeks the motion vector of the key position pixel of human skeleton, extracts the pixel wherein moved, and carries out moving region and draw
Point；
The type of sports categorization module, for according in every figure moving region quantity and direction to type of sports into
Row classification；
The motor message statistical module, for calculating the statistics of the motor message of picture under each operation type is classified
Amount；
Described image prediction module, for being input to trained SVM model for the statistic of motor message as feature
The middle prediction for carrying out act of violence.
As advanced optimizing, the Acquiring motion area module carries out moving region using k-means clustering method
It divides.
As advanced optimizing, type of sports is divided by the type of sports categorization module: single moving region, do more physical exercises area
Domain in the same direction and is done more physical exercises region three classes not in the same direction.
As advanced optimizing, the statistic for the motor message that the motor message statistical module calculates includes: to count respectively
Calculation amplitude, acceleration, the maximum value of the degree of association, minimum value, mean value, intermediate value and standard deviation.
As advanced optimizing, the Acquiring motion area module, type of sports categorization module, motor message statistical module
It is deployed in the same server with image prediction module；Alternatively, Acquiring motion area module, type of sports categorization module
It is deployed in the same server with motor message statistical module, described image prediction module is deployed in another server
In.
In addition, the present invention also provides a kind of act of violence detection method based on human skeleton and motor message feature,
Itself the following steps are included:
A. 3 frames are continuously extracted by certain time interval, and the key of human skeleton is then sought according to the frame difference of adjacent two frame
The motion vector of position pixel extracts the pixel wherein moved, and carries out moving region division；
B. according in every figure moving region quantity and direction classify to type of sports；
C. the statistic of the motor message of picture under each operation type is classified is calculated；
D. the statistic of motor message is input in trained SVM model as feature and carries out the pre- of act of violence
It surveys.
As advanced optimizing, in step a, the division of moving region is carried out using k-means clustering method.
As advanced optimizing, in step b, the type of sports is divided into: single moving region, region of doing more physical exercises are in the same direction and more
Moving region three classes not in the same direction.
As advanced optimizing, in step c, the calculation method of the statistic of the motor message includes:
According to formulaCalculate the amplitude M of figure；Wherein, r is region, nrIndicate the area r
Number of pixels in domain, x and y respectively indicate the cross of pixel, ordinate；
According to formula A (t)=| M (t)-M (t-1) |, calculate the acceleration A of figure；
According to formulaCalculate the degree of association G in two regions in free hand drawing；Wherein, r is region, Dij
Indicate the region i at a distance from j regional center.
As advanced optimizing, in step c, the statistic of motor message includes: the maximum of amplitude, acceleration, the degree of association
Value, minimum value, mean value, intermediate value and standard deviation.
The beneficial effects of the present invention are:
Based on human skeleton and motor message feature, motion analysis is carried out to the video flowing of input, monitors and fights
It when incident of violence, immediately triggers warning device and alarms, administrative staff can then be notified at the first time, and be carried out corresponding
Processing, have real-time；
Due to be based on human skeleton, to human body key position carry out motion analysis, and the motion amplitude of movement pixel,
Acceleration, three kinds of motor messages of the region degree of association statistic carry out forecast analysis as final characteristic information, to improve pre-
The accuracy of survey.
Detailed description of the invention
Fig. 1 is the act of violence detection method flow chart in the present invention；
Fig. 2 is the act of violence detection system in embodiment 1 in centralized processing scene application schematic diagram；
Fig. 3 is the act of violence detection system in embodiment 2 in distributed treatment scene application schematic diagram.
Specific embodiment
The present invention is directed to propose a kind of act of violence detection system and method based on human skeleton and motor message feature,
The real-time and accuracy of detection are improved, and adapts to a variety of different monitoring environment.It passes through one section of short video sequences of input,
Successively detect whether to occur by Acquiring motion area, type of sports classification, calculating motor message statistic and image prediction
Incident of violence.Wherein, in Acquiring motion area, it is based on human skeleton, motion analysis is carried out to human body key position, is being calculated
When motor message statistic, the statistic of three kinds of motion amplitude of movement pixel, acceleration, region degree of association motor messages is made
It is input in SVM model for final characteristic information.
For ease of understanding, the technical term being likely to occur in the present invention is explained first:
1. video frame: what video was all made of static picture, these static pictures are referred to as frame.In general,
Frame per second is lower than 15 frames/second, and continuous sport video just has the feeling of pause.Using television standard PAL system, it is advised in China
Determine 25 frames of video/second (interlace mode), 625 scan lines of every frame.Frame number is more, and data volume is bigger, so sometimes for subtracting
Lack data volume and slowed down frame speed, such as only 16 frames are per second, can reach certain satisfaction, but effect is slightly poor.
2. human skeleton: human body shares 206 pieces of bones, is divided into 3 skull, bone of body and Limb bone major parts.Wherein, have
29 pieces of skull, 51 pieces of bone of body, 126 pieces of Limb bone.
3. motion amplitude: refer to body or certain a part certain distance between reference or angle (can with the degree of measurement,
Scale and line segment unit are determined or are indicated) between the value that moves.The space characteristics that it is showed then are decided by athletic performance
Task and requirement, the quantity of muscle group, the elasticity of muscle fibre and the flexibility in joint.
The act of violence detection system based on human skeleton and motor message feature in the present invention, comprising: moving region
Extraction module, type of sports categorization module, motor message statistical module and image prediction module；
The Acquiring motion area module, for continuously extracting 3 frames by certain time interval；Then according to adjacent two frame
Frame difference seeks the motion vector of the key position pixel of human skeleton, extracts the pixel wherein moved, and carries out moving region and draw
Point；
The type of sports categorization module, for according in every figure moving region quantity and direction to type of sports into
Row classification；
The motor message statistical module, for calculating the statistics of the motor message of picture under each operation type is classified
Amount；
Described image prediction module, for being input to trained SVM model for the statistic of motor message as feature
The middle prediction for carrying out act of violence.
In specific deployment, the Acquiring motion area module, type of sports categorization module, motor message statistical module and
Image prediction module can be deployed in the same server；Alternatively, Acquiring motion area module, type of sports categorization module
It is deployed in the same server with motor message statistical module, image prediction module is deployed in another server.
Based on said detecting system, the detection method that the present invention realizes is as shown in Figure 1 comprising following implemented step:
1, short video sequences S is inputted；
In this step, monitor video of the short video sequences from the analysis to be detected of camera shooting.
2, three frames are continuously extracted:
In this step, three frame image F1, F2, F3 can be continuously extracted according to the frame period of setting.
3, human skeleton key position motion vector is calculated:
In this step, the key position pixel of human skeleton can be sought according to frame difference F1F2, F2F3 of adjacent two frame
Motion vector O1 and O2.
4, movement pixel and moving region are extracted:
In this step, the pixel of the movement in motion vector is extracted, and moving region is carried out using k-means clustering method
It divides.
5, type of sports classification is carried out to free hand drawing:
In this step, schemes for each, classified according to the number of moving region and direction, Dan Yun can be divided into
Region is done more physical exercises in the same direction and region of doing more physical exercises three classes not in the same direction in dynamic region.
6, the motor message statistic of the picture of each classification is calculated:
In this step, the amplitude of picture of each classification, acceleration, the maximum value of the degree of association, minimum value, are calculated
Value, intermediate value and standard deviation.
7, feature input SVM is predicted:
In this step, motor message statistic is input in preparatory trained SVM model as characteristic quantity and is carried out in advance
It surveys, and exports result: if SVM prediction result is 1, then it represents that video is the video containing act of violence, if SVM prediction result is 0,
Then indicate that video is the video without act of violence.
Embodiment 1:
The scene of the present embodiment description is: four modules are all deployed on detection service device, and camera is responsible for always video
Prediction result is periodically stored in database by the acquisition of stream, and client is then gone to check whether there is by the way of poll sudden and violent
The generation of power event.Referring to Fig. 2, specific steps include:
Camera acquires video flowing in real time first, and is buffered in local.
Module one (Acquiring motion area module) in detection service device carries out the motor image of the key position of human skeleton
The extraction of element and moving region, the key step of the module have:
Three frame F1, F2, F3 are continuously extracted according to pre-set frame period；
The motion vector set O1 that the critical movements characteristic point of human skeleton is obtained out by F1F2, is obtained in place by F2F3
Shift to duration set O2；
Because motion vector is not 0, then the pixel is movement pixel.Movement pixel therein is extracted, and uses cluster side
Method will move pixel and carry out moving region division；
Module two (type of sports categorization module) classifies figure according to the number and direction of moving region: a Dan Yun
Dynamic region, two do more physical exercises region in the same direction, two regions of doing more physical exercises it is reversed.
Module three (motor message statistical module) calculates separately motor message, the i.e. maximum of amplitude, acceleration, the degree of association
Value, minimum value, intermediate value, standard deviation；
According to formulaWherein, r is region, nrIndicate number of pixels in the region r, meter
Calculate the amplitude M of figure；
According to formula A (t)=| M (t)-M (t-1) |, calculate the acceleration A of figure；
According to formulaWherein, r is region, DijIndicate the region i at a distance from j regional center,
Calculate the degree of association G in two regions in free hand drawing；
Amplitude, acceleration, the maximum value of the degree of association, minimum value, mean value, the intermediate value, standard deviation for calculating separately video flowing, obtain
To final feature V.
Module four (image prediction module) is using the motor message statistical vector V of acquisition as feature vector, by instructing in advance
The SVM model perfected is predicted, and exports corresponding prediction result.SVM is trained by great amount of samples in advance.
After above-mentioned a series of processing step, the incident of violence that detection service device will test is stored in database
In.In this embodiment, since four modules are all deployed in the same server, processing speed is relatively slow, is suitable for one
It is not very high public place to requirement of real-time a bit, such as: street etc..
Last client will inquire database by the way of poll, and whether detection occurs act of violence in a period of time.
If detecting act of violence, the operator on duty for notifying this area is gone to corresponding place to check by administrator.
Embodiment 2:
The scene of the present embodiment description is: module one, module two and module three are all deployed on same video stream server,
And module four is deployed on individual detection service device.Camera is responsible for always the acquisition of video flowing, and video stream server is main
It is to complete first three module, deals with than relatively time-consuming, therefore individually deployment.Detection service device is then concentrated and is predicted, in this way
Time-consuming module can be separately separated out, to improve the processing capacity of whole system.Last client uses the side of poll
Formula goes to check whether there is the generation of incident of violence.Referring to Fig. 3, the specific steps are as follows:
Camera acquires video flowing in real time first, and is buffered in local.
Module one (Acquiring motion area module) in video stream server carries out the extraction of movement pixel and moving region,
The key step of the module has:
Three frame F1, F2, F3 are continuously extracted according to pre-set frame period；
The motion vector set O1 that the critical movements characteristic point of human skeleton is obtained out by F1F2, is obtained in place by F2F3
Shift to duration set O2；
Because motion vector is not 0, then the pixel is movement pixel.Movement pixel therein is extracted, and uses cluster side
Method will move pixel and carry out moving region division；
Module two (type of sports categorization module) classifies free hand drawing according to the number and direction of moving region: a fortune
Dynamic region, two moving regions are in the same direction, two moving regions are not in the same direction.
Module three (motor message statistical module) calculates separately motor message, the i.e. maximum of amplitude, acceleration, the degree of association
Value, minimum value, intermediate value, standard deviation；
According to formulaWherein, r is region, nrIndicate number of pixels in the region r, meter
Calculate the amplitude M of free hand drawing；
According to formula A (t)=| M (t)-M (t-1) |, calculate the acceleration A of free hand drawing；
According to formulaWherein, r is region, DijIndicate the region i at a distance from j regional center,
Calculate the degree of association G in two regions in free hand drawing；
Amplitude, acceleration, the maximum value of the degree of association, minimum value, mean value, the intermediate value, standard deviation for calculating separately video flowing, obtain
To final characteristic set V.
After characteristic set V is calculated in video stream server, it is uploaded to detection service device immediately；
Detection service device will receive the characteristic set V from multiple video stream servers, and then using module four, (image is pre-
Survey module) it is predicted by trained SVM model, and export corresponding prediction result.SVM passes through great amount of samples in advance
It is trained.
After above-mentioned a series of processing step, the incident of violence that detection service device will test is stored in database
In.In this embodiment, it is disposed since the first two time-consuming module is separately separated out, is able to ascend the processing energy of whole system
Power, the Code in Hazardous Special Locations high suitable for some pairs of requirement of real-time, such as: detention house etc..
Last client will inquire database by the way of poll, and whether detection occurs act of violence in a period of time.
If detecting act of violence, the operator on duty for notifying this area is gone to corresponding place to check by administrator.
In above two example scheme, database is to the storage mode of testing result referring to following table:
Act of violence based on human skeleton and motor message feature detects storage table
There are six fields in total for the table, are respectively as follows:
Id, i.e. serial number, the number of the video flowing to label detection；
Timestamp, i.e. timestamp, to identify the time for carrying out act of violence detection to video flowing；
RoomID, the i.e. serial number of warehouse, to identify the number in region or room locating for camera；
CameraId, i.e. camera ID, the number of the camera to identify video flowing source；
MotionSig_statistic, i.e. motor message statistic, by the motor message counting statistics of human skeleton feature
It obtains, to the input as prediction model；
Whether isViolent is violence, to identify the act of violence testing result of video flowing, indicate video flowing for 1
Containing act of violence, indicate that video flowing is free of act of violence for 0.
The video of some corresponding camera is inquired according to roomID and cameraId in the above-mentioned storage table of clients poll
The act of violence testing result of stream, according to the value of isViolent field to determine whether there are act of violence, if isViolent
The value of field, which is 1, indicates act of violence, and by the corresponding warning device of automatic trigger, administrator can notify the value of corresponding area
Class personnel go to corresponding location to check and handle.
Since above-mentioned storage table has recorded the motor message statistic MotionSig_ calculated every time for video flowing
Statistic and act of violence testing result isViolent, our SVM model can be extracted from database largely to be gone through
History data are trained, that is, the training sample for SVM model of enriching constantly, to keep SVM prediction more and more accurate.
Claims (10)
1. the act of violence detection system based on human skeleton and motor message feature, which is characterized in that
It include: Acquiring motion area module, type of sports categorization module, motor message statistical module and image prediction module；
The Acquiring motion area module, for continuously extracting 3 frames by certain time interval；Then the frame according to adjacent two frame is poor
The motion vector of the key position pixel of human skeleton is sought, extracts the pixel wherein moved, and carry out moving region division；
The type of sports categorization module, for according in every figure moving region quantity and direction type of sports is divided
Class；
The motor message statistical module, for calculating the statistic of the motor message of picture under each operation type is classified；
Described image prediction module, for using the statistic of motor message as feature be input in trained SVM model into
The prediction of row act of violence.
2. the act of violence detection system based on human skeleton and motor message feature, feature exist as described in claim 1
In the Acquiring motion area module carries out the division of moving region using k-means clustering method.
3. the act of violence detection system based on human skeleton and motor message feature, feature exist as described in claim 1
In type of sports is divided by the type of sports categorization module: single moving region, region of doing more physical exercises is in the same direction and does more physical exercises region not
Three classes in the same direction.
4. the act of violence detection system based on human skeleton and motor message feature, feature exist as described in claim 1
In the statistic for the motor message that the motor message statistical module calculates includes: the amplitude that calculates separately, acceleration, the degree of association
Maximum value, minimum value, mean value, intermediate value and standard deviation.
5. the act of violence detection system based on human skeleton and motor message feature as described in claim 1-4 any one
System, which is characterized in that the Acquiring motion area module, type of sports categorization module, motor message statistical module and image are pre-
Module is surveyed to be deployed in the same server；Alternatively, Acquiring motion area module, type of sports categorization module and movement letter
Number statistical module is deployed in the same server, and described image prediction module is deployed in another server.
6. the act of violence detection method based on human skeleton and motor message feature, which comprises the following steps:
A. 3 frames are continuously extracted by certain time interval, and the key position of human skeleton is then sought according to the frame difference of adjacent two frame
The motion vector of pixel extracts the pixel wherein moved, and carries out moving region division；
B. according in every figure moving region quantity and direction classify to type of sports；
C. the statistic of the motor message of picture under each operation type is classified is calculated；
D., the statistic of motor message is input to the prediction that act of violence is carried out in trained SVM model as feature.
7. the act of violence detection method based on human skeleton and motor message feature, feature exist as claimed in claim 6
In in step a, using the division of k-means clustering method progress moving region.
8. the act of violence detection method based on human skeleton and motor message feature, feature exist as claimed in claim 6
In in step b, the type of sports is divided into: single moving region, do more physical exercises region in the same direction and region of doing more physical exercises three classes not in the same direction.
9. the act of violence detection method based on human skeleton and motor message feature, feature exist as claimed in claim 6
In in step c, the calculation method of the statistic of the motor message includes:
According to formulaCalculate the amplitude M of figure；Wherein, r is region, nrIt indicates in the region r
Number of pixels, x and y respectively indicate the cross of pixel, ordinate；
According to formula A (t)=| M (t)-M (t-1) |, calculate the acceleration A of figure；
According to formulaCalculate the degree of association G in two regions in free hand drawing；Wherein, r is region, DijIt indicates
The region i is at a distance from j regional center.
10. the act of violence detection method based on human skeleton and motor message feature, feature exist as claimed in claim 9
In, in step c, the statistic of motor message include: amplitude, acceleration, the maximum value of the degree of association, minimum value, mean value, intermediate value and
Standard deviation.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810864496.5A CN109086717B (en) | 2018-08-01 | 2018-08-01 | Violent behavior detection system and method based on human skeleton and motion signal characteristics |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810864496.5A CN109086717B (en) | 2018-08-01 | 2018-08-01 | Violent behavior detection system and method based on human skeleton and motion signal characteristics |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109086717A true CN109086717A (en) | 2018-12-25 |
| CN109086717B CN109086717B (en) | 2022-05-24 |
Family
ID=64831267
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810864496.5A Active CN109086717B (en) | 2018-08-01 | 2018-08-01 | Violent behavior detection system and method based on human skeleton and motion signal characteristics |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109086717B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110274590A (en) * | 2019-07-08 | 2019-09-24 | 哈尔滨工业大学 | A kind of violent action detection method and system based on decision tree |
| CN110490148A (en) * | 2019-08-22 | 2019-11-22 | 四川自由健信息科技有限公司 | A kind of recognition methods for behavior of fighting |
| WO2022193412A1 (en) * | 2021-03-15 | 2022-09-22 | 华南理工大学 | Video violence identification method and system based on human skeleton point cloud interaction learning, and medium |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20050152588A1 (en) * | 2003-10-28 | 2005-07-14 | University Of Chicago | Method for virtual endoscopic visualization of the colon by shape-scale signatures, centerlining, and computerized detection of masses |
| 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 |
| CN103150579A (en) * | 2013-02-25 | 2013-06-12 | 东华大学 | Abnormal human behavior detecting method based on video sequence |
| US9355306B2 (en) * | 2013-09-27 | 2016-05-31 | Konica Minolta Laboratory U.S.A., Inc. | Method and system for recognition of abnormal behavior |
| CN106980829A (en) * | 2017-03-17 | 2017-07-25 | 苏州大学 | Abnormal behaviour automatic testing method of fighting based on video analysis |
-
2018
- 2018-08-01 CN CN201810864496.5A patent/CN109086717B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20050152588A1 (en) * | 2003-10-28 | 2005-07-14 | University Of Chicago | Method for virtual endoscopic visualization of the colon by shape-scale signatures, centerlining, and computerized detection of masses |
| 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 |
| CN103150579A (en) * | 2013-02-25 | 2013-06-12 | 东华大学 | Abnormal human behavior detecting method based on video sequence |
| US9355306B2 (en) * | 2013-09-27 | 2016-05-31 | Konica Minolta Laboratory U.S.A., Inc. | Method and system for recognition of abnormal behavior |
| CN106980829A (en) * | 2017-03-17 | 2017-07-25 | 苏州大学 | Abnormal behaviour automatic testing method of fighting based on video analysis |
Non-Patent Citations (4)
| Title |
|---|
| CHUNYU CHEN 等: "Detection of Anomalous Crowd Behavior Based on the Acceleration Feature", 《IEEE SENSORS JOURNAL》 * |
| 秦陈刚: "面向监视场景的斗殴行为检测技术研究", 《中国优秀硕士学位论文全文数据库信息科技辑》 * |
| 许莙苓: "基于光流特征的航站楼旅客异常行为识别方法研究", 《中国优秀硕士学位论文全文数据库工程科技II辑》 * |
| 郁映卓: "基于人体运动特征的异常行为检测和姿态识别", 《中国优秀硕士学位论文全文数据库信息科技辑》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110274590A (en) * | 2019-07-08 | 2019-09-24 | 哈尔滨工业大学 | A kind of violent action detection method and system based on decision tree |
| CN110490148A (en) * | 2019-08-22 | 2019-11-22 | 四川自由健信息科技有限公司 | A kind of recognition methods for behavior of fighting |
| WO2022193412A1 (en) * | 2021-03-15 | 2022-09-22 | 华南理工大学 | Video violence identification method and system based on human skeleton point cloud interaction learning, and medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109086717B (en) | 2022-05-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102542492B (en) | System and method for evaluating effect of visual advertisement | |
| CN106998444B (en) | Big data face monitoring system | |
| Javed et al. | Knight/spl trade: a real time surveillance system for multiple and non-overlapping cameras | |
| CN105405150B (en) | Anomaly detection method and device based on fusion feature | |
| CN109086717A (en) | Act of violence detection system and method based on human skeleton and motor message feature | |
| CN107241572B (en) | Training video tracking evaluation system for students | |
| CN108647612A (en) | Billboard watches flow of the people analysis system | |
| CN103902966B (en) | Video interactive affair analytical method and device based on sequence space-time cube feature | |
| CN112131927B (en) | Sow delivery time prediction system based on gesture conversion characteristics in late gestation period | |
| CN110135883A (en) | A kind of method and system of elevator crowd portrayal and commercial audience analysis | |
| CN109886165A (en) | A kind of action video extraction and classification method based on moving object detection | |
| CN109117771A (en) | Incident of violence detection system and method in a kind of image based on anchor node | |
| Arif et al. | Counting of people in the extremely dense crowd using genetic algorithm and blobs counting | |
| CN103514429B (en) | Detect the method and image processing equipment of the privileged site of object | |
| CN107330414A (en) | Act of violence monitoring method | |
| CN113963315A (en) | Real-time video multi-user behavior recognition method and system in complex scene | |
| CN112329691A (en) | Monitoring video analysis method and device, electronic equipment and storage medium | |
| CN110287907A (en) | A kind of method for checking object and device | |
| CN106709438A (en) | Method for collecting statistics of number of people based on video conference | |
| CN109063667A (en) | A kind of video identification method optimizing and method for pushing based on scene | |
| CN108875712A (en) | A kind of act of violence detection system and method based on ViF descriptor | |
| CN102609719A (en) | Method for identifying place image on the basis of improved probabilistic topic model | |
| CN116229518B (en) | Bird species observation method and system based on machine learning | |
| CN109918996A (en) | The illegal action identification method of personnel, system, computer equipment and storage medium | |
| CN111860457A (en) | Fighting behavior recognition early warning method and recognition early warning system thereof |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |