CN106157325B - Group abnormal behavior detection method and system - Google Patents
Group abnormal behavior detection method and system Download PDFInfo
- Publication number
- CN106157325B CN106157325B CN201510160419.8A CN201510160419A CN106157325B CN 106157325 B CN106157325 B CN 106157325B CN 201510160419 A CN201510160419 A CN 201510160419A CN 106157325 B CN106157325 B CN 106157325B
- Authority
- CN
- China
- Prior art keywords
- group
- cooperativity
- value
- tracking
- feature points
- 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.)
- Active
Links
Images
Landscapes
- Image Analysis (AREA)
- Investigating Or Analysing Materials By Optical Means (AREA)
- Closed-Circuit Television Systems (AREA)
Abstract
The invention relates to a group abnormal behavior detection method, which comprises the following steps: performing group tracking on the groups in the video frame by adopting a group tracking method of dynamic evolution of group structures; in the group tracking process, recording and storing the group cooperativity value of each frame of the video frame, which is tracked to the group, until the group tracking is finished; and calculating the group cooperativity value in the d frame of the group according to the group cooperativity value of each frame of the tracked group, and judging whether abnormal behaviors occur or not according to a preset threshold value. The invention also relates to a group abnormal behavior detection system. The method directly utilizes the basic characteristics of the group to detect the abnormal behavior, is very simple and convenient, avoids the complex model learning process, and improves the efficiency of searching the reason causing the safety problem.
Description
Technical Field
The invention relates to a group abnormal behavior detection method and system.
Background
In the modern society with rapid development, the population grows faster and faster, and the problems caused by a series of populations are more prominent. The safety problem is particularly prominent in places with relatively dense crowd density, such as railway stations, bus stations and the like. Through the detection to unusual action, can filter a large amount of information useless to the security protection in the surveillance video, practiced thrift a large amount of manpowers. In a public place with complicated personnel, the monitoring video is analyzed, and then the occurrence of some events causing safety problems is judged and corresponding responses are made in time, so that not only can emergency events be effectively processed, but also outstanding contributions are made to the aspects of maintaining the safety of the public place, the safety of lives and properties of people and the like.
At present, due to the large population size and density, the abnormal behaviors of the population are mostly studied in a macroscopic view, namely, the population is studied as a whole. Mainly comprises the following steps: detecting and tracking a video moving target; monitoring according to the movement characteristics of the population; and modeling the group track through the model, and identifying the abnormal behavior of the group.
Therefore, most of the existing group abnormal behavior detection modes need to establish a model and then carry out model learning, so that the efficiency is low and the process is complex.
Disclosure of Invention
In view of the above, it is desirable to provide a group abnormal behavior detection method and system.
The invention provides a group abnormal behavior detection method, which comprises the following steps: a. performing group tracking on the groups in the video frame by adopting a group tracking method of dynamic evolution of group structures; b. in the group tracking process, recording and storing the group cooperativity value of each frame of the video frame, which is tracked to the group, until the group tracking is finished; c. and calculating the group cooperativity value in the d frame of the group according to the group cooperativity value of each frame of the tracked group, and judging whether abnormal behaviors occur or not according to a preset threshold value.
Wherein, the step a specifically comprises: tracking the feature points extracted from the video frame by using an optical flow method, and obtaining the motion information of the feature points; calculating whether the motion modes of the feature points are consistent or not according to the obtained motion information of the feature points, and clustering the feature points with consistent motion modes according to the density to ensure that the density of the feature points contained in the generated block is higher; detecting and obtaining the group in the video frame according to the generated block by adopting a group merging method; and carrying out group tracking on the groups in the video frames obtained by the detection by adopting a layered dynamic tree structure.
The population synergy value is obtained by: detecting a block composed of feature points; obtaining a group cooperative motion mode of each feature point; obtaining the group cooperativity motion mode of the block according to the group cooperativity motion mode of most feature points contained in the block; and obtaining the group cooperativity value of the group according to the group cooperativity movement mode of the blocks.
The calculating the group cooperativity value in the group d frame in the step c specifically includes: and calculating the group cooperativity value of the group in the d frame by a function phi every d frames, wherein the function phi can carry out average calculation or variance calculation on the group cooperativity value of the group in the d frames.
The step c of judging whether abnormal behaviors occur or not according to a preset threshold specifically comprises the following steps: making a difference with the phi value of the previous d frame; when the difference value is larger than a preset threshold value T, judging that an abnormality occurs; if the difference is smaller than the threshold T, repeating the process until the group tracking is finished, and if the difference is still smaller than the threshold T until the group tracking is finished, judging that the group has no abnormal behavior.
The invention also provides a group abnormal behavior detection system, which comprises a tracking module, a storage module and a judgment module, wherein: the tracking module is used for tracking the groups in the video frame by adopting a group tracking method of dynamic evolution of group structures; the storage module is used for recording and storing the group cooperativity value of each frame of the video frames, which is tracked to the group, in the group tracking process until the group tracking is finished; the judging module is used for calculating the group cooperativity value in the d frame of the group according to the group cooperativity value of each frame of the tracked group and judging whether abnormal behaviors occur or not according to a preset threshold value.
Wherein the tracking module is specifically configured to: tracking the feature points extracted from the video frame by using an optical flow method, and obtaining the motion information of the feature points; calculating whether the motion modes of the feature points are consistent or not according to the obtained motion information of the feature points, and clustering the feature points with consistent motion modes according to the density to ensure that the density of the feature points contained in the generated block is higher; detecting and obtaining the group in the video frame according to the generated block by adopting a group merging method; and carrying out group tracking on the groups in the video frames obtained by the detection by adopting a layered dynamic tree structure.
The group cooperativity value is obtained by the following method: detecting a block composed of feature points; obtaining a group cooperative motion mode of each feature point; obtaining the group cooperativity motion mode of the block according to the group cooperativity motion mode of most feature points contained in the block; and obtaining the group cooperativity value of the group according to the group cooperativity movement mode of the blocks.
The judgment module is specifically configured to: calculating the group cooperativity value of the group in the d frames by a function phi every d frames; making a difference with the phi value of the previous d frame; when the difference value is larger than a preset threshold value T, judging that an abnormality occurs; if the difference is smaller than the threshold T, repeating the process until the group tracking is finished, and if the difference is still smaller than the threshold T until the group tracking is finished, judging that the group has no abnormal behavior.
The function phi can perform average calculation on the group cooperativity values of the groups in the d frames or perform variance calculation.
The method and the system for detecting the abnormal behaviors of the group directly utilize some basic characteristics of the group to detect the abnormal behaviors, are very simple and convenient, and avoid a complex model learning process. The invention can detect the event that the group movement suddenly becomes disordered, not only can improve the efficiency of searching the reason causing the safety problem, but also can lead the related personnel to make corresponding response in time and effectively process the emergency.
Drawings
FIG. 1 is a flow chart of a method for detecting abnormal group behaviors according to the present invention;
fig. 2 is a diagram of a hardware architecture of a group abnormal behavior detection system according to the present invention.
Detailed Description
The present invention will be described in further detail with reference to the accompanying drawings and specific embodiments.
Fig. 1 is a flowchart illustrating an operation of a group abnormal behavior detection method according to a preferred embodiment of the present invention.
And step S1, performing group tracking on the groups in the video frame by adopting a group tracking method of dynamic evolution of the group structure. The method comprises the following specific steps:
(1) and tracking the feature points extracted from the video frame by using an optical flow method, and obtaining the motion information of the feature points.
(2) And calculating whether the motion patterns of the feature points are consistent or not according to the obtained motion information of the feature points, and clustering the feature points with consistent motion patterns according to density, namely generating blocks (patch) by classes with high feature point density.
(3) And detecting the groups in the video frame according to the generated blocks by adopting a group Merging (collecting) method.
In the present embodiment, when detecting a population, population detection is performed by a group-cooperative motion pattern. In particular, because each population has one population cooperative motion pattern, different populations can be distinguished by the population cooperative motion pattern. The invention is based on this to detect abnormal behavior.
(4) And carrying out group tracking on the groups in the video frames obtained by the detection by adopting a layered dynamic tree structure.
Step S2, in the process of group tracking, record and store the group cooperativity value of each frame of the video frames where the group is tracked until the group tracking is finished. The method comprises the following specific steps:
when the group cooperativity value is stored, an ID number is allocated to each tracked group in the video frame, and then the group cooperativity value of the group corresponding to the ID number in each frame is recorded and stored. Wherein the ID number assigned by the population remains unchanged throughout the tracking process.
Population synergy (crowd collectibility): population synergy measures the degree to which each individual in a population is synergistic in a motor pattern. The group synergy value is relatively high if the degree of group synergy is greater, and is relatively low otherwise. The population synergy value is obtained by:
in the embodiment, when detecting a group, a block composed of feature points is detected first; the tracking of the characteristic points can obtain the motion information of the characteristic points, so that the group cooperative motion mode of each characteristic point is obtained; and finally, detecting and obtaining the group in the video frame according to the generated block by adopting a group combination method, and obtaining the group cooperativity value of the group according to the group cooperativity motion mode of the block.
And step S3, calculating the group cooperativity value of the group within a certain time period according to the group cooperativity value of each frame of the tracked group, and judging whether abnormal behaviors occur or not according to a preset threshold value. Specifically, the method comprises the following steps:
each tracked group has a group synergy (Crowd collectibility) value, and when the pedestrian movement in the group is relatively consistent and stable, the group synergy value is relatively stable and has a higher value. When the tracked group conflicts with other groups or the movement of pedestrians in the group is disordered due to other reasons, the group cooperativity value is suddenly reduced. When the sudden reduction of the group cooperativity value is detected, the abnormal behavior of the group can be judged.
The method comprises the following specific steps:
and calculating the group cooperativity value of the group in the d frame by using a function phi every d frames, wherein in the specific calculation, the function phi can carry out average calculation or variance calculation on the group cooperativity value of the group in the d frames. Then making a difference with the phi value of the previous d frame, when the difference value is larger than a preset threshold value T, namely phin-φn-1>And T, judging that the abnormity occurs. If the difference is smaller than the preset threshold value T, repeating the process until the group tracking is finished. And if the difference is still smaller than the preset threshold value until the whole tracking process of the group is finished, judging that the group has no abnormal behavior.
Fig. 2 is a diagram showing a hardware architecture of a group abnormal behavior detection system according to the present invention. The system comprises a tracking module, a storage module and a judgment module.
The tracking module is used for tracking the groups in the video frame by adopting a group tracking method of dynamic evolution of group structures. The method comprises the following specific steps:
(1) and tracking the feature points extracted from the video frame by using an optical flow method, and obtaining the motion information of the feature points.
(2) And calculating whether the motion patterns of the feature points are consistent or not according to the obtained motion information of the feature points, and clustering the feature points with consistent motion patterns according to density, namely generating blocks (patch) by classes with high feature point density.
(3) And detecting the groups in the video frame according to the generated blocks by adopting a group Merging (collecting) method.
In the present embodiment, when detecting a population, population detection is performed by a group-cooperative motion pattern. In particular, because each population has one population cooperative motion pattern, different populations can be distinguished by the population cooperative motion pattern. The invention is based on this to detect abnormal behavior.
(4) And carrying out group tracking on the groups in the video frames obtained by the detection by adopting a layered dynamic tree structure.
The storage module is used for recording and storing the group cooperativity value of each frame of the video frames, which is tracked to the group, in the group tracking process until the group tracking is finished. The method comprises the following specific steps:
when the group cooperativity value is stored, an ID number is allocated to each tracked group in the video frame, and then the group cooperativity value of the group corresponding to the ID number in each frame is recorded and stored. Wherein the ID number assigned by the population remains unchanged throughout the tracking process.
The population synergy value is obtained by:
in the embodiment, when detecting a group, a block composed of feature points is detected first; the tracking of the characteristic points can obtain the motion information of the characteristic points, so that the group cooperative motion mode of each characteristic point is obtained; and finally, detecting and obtaining the group in the video frame according to the generated block by adopting a group combination method, and obtaining the group cooperativity value of the group according to the group cooperativity motion mode of the block.
The judging module is used for calculating the group cooperativity value of the group within a certain time period according to the group cooperativity value of each frame of the tracked group and judging whether abnormal behaviors occur or not according to a preset threshold value. Specifically, the method comprises the following steps:
each tracked group has a group synergy (Crowd collectibility) value, and when the pedestrian movement in the group is relatively stable, the group synergy value is relatively stable and has a higher value. When the tracked group conflicts with other groups or the movement of pedestrians in the group is disordered due to other reasons, the group cooperativity value is suddenly reduced. When the sudden reduction of the group cooperativity value is detected, the abnormal behavior of the group can be judged.
The method comprises the following specific steps:
and calculating the group cooperativity value of the group in the d frame by using a function phi every d frames, wherein in the specific calculation, the function phi can carry out average calculation or variance calculation on the group cooperativity value of the group in the d frames. Then making a difference with the phi value of the previous d frame, when the difference value is larger than a preset threshold value T, namely phin-φn-1>And T, judging that the abnormity occurs. If the difference is smaller than the preset threshold value T, repeating the process until the group tracking is finished. And if the difference is still smaller than the preset threshold value until the whole tracking process of the group is finished, judging that the group has no abnormal behavior.
Although the present invention has been described with reference to the presently preferred embodiments, it will be understood by those skilled in the art that the foregoing description is illustrative only and is not intended to limit the scope of the invention, as claimed.
Claims (4)
1. A method for detecting abnormal behavior of a population, the method comprising:
a, performing group tracking on groups in a video frame by adopting a group tracking method of dynamic evolution of a group structure;
b, in the group tracking process, recording and storing the group cooperativity value of each frame of the tracked group in the video frame until the group tracking is finished;
c, calculating the group cooperativity value in the d frame of the group according to the group cooperativity value of each frame of the tracked group, and judging whether abnormal behaviors occur or not according to a preset threshold value;
the step a specifically comprises the following steps:
tracking the feature points extracted from the video frame by using an optical flow method, and obtaining the motion information of the feature points;
calculating whether the motion modes of the feature points are consistent or not according to the obtained motion information of the feature points, and clustering the feature points with consistent motion modes according to the density to ensure that the density of the feature points contained in the generated block is higher;
detecting and obtaining the group in the video frame according to the generated block by adopting a group merging method;
performing group tracking on the groups in the video frames obtained by the detection by adopting a layered dynamic tree structure;
the population synergy value is obtained by:
detecting a block composed of feature points; obtaining a group cooperative motion mode of each feature point; obtaining the group cooperativity motion mode of the block according to the group cooperativity motion mode of most feature points contained in the block; obtaining a group cooperativity value of the group according to the group cooperativity movement mode of the block;
the step c of judging whether abnormal behaviors occur or not according to a preset threshold specifically comprises the following steps:
making a difference with the phi value of the previous d frame;
when the difference value is larger than a preset threshold value T, judging that an abnormality occurs;
if the difference is smaller than the threshold T, repeating the process until the group tracking is finished, and if the difference is still smaller than the threshold T until the group tracking is finished, judging that the group has no abnormal behavior.
2. The method of claim 1, wherein the calculating of the group cooperativity values within the group d frames in step c comprises:
and calculating the group cooperativity value of the group in the d frame by a function phi every d frames, wherein the function phi can carry out average calculation or variance calculation on the group cooperativity value of the group in the d frames.
3. The group abnormal behavior detection system is characterized by comprising a tracking module, a storage module and a judgment module, wherein:
the tracking module is used for tracking the groups in the video frame by adopting a group tracking method of dynamic evolution of group structures;
the storage module is used for recording and storing the group cooperativity value of each frame of the video frames, which is tracked to the group, in the group tracking process until the group tracking is finished;
the judging module is used for calculating the group cooperativity value in the d frame of the group according to the group cooperativity value of each frame of the tracked group and judging whether abnormal behaviors occur or not according to a preset threshold value;
the tracking module is specifically configured to:
tracking the feature points extracted from the video frame by using an optical flow method, and obtaining the motion information of the feature points;
calculating whether the motion modes of the feature points are consistent or not according to the obtained motion information of the feature points, and clustering the feature points with consistent motion modes according to the density to ensure that the density of the feature points contained in the generated block is higher;
detecting and obtaining the group in the video frame according to the generated block by adopting a group merging method;
performing group tracking on the groups in the video frames obtained by the detection by adopting a layered dynamic tree structure;
the group cooperativity value is obtained by the following method:
detecting a block composed of feature points; obtaining a group cooperative motion mode of each feature point; obtaining the group cooperativity motion mode of the block according to the group cooperativity motion mode of most feature points contained in the block; obtaining a group cooperativity value of the group according to the group cooperativity movement mode of the block;
the judgment module is specifically configured to:
calculating the group cooperativity value of the group in the d frames by a function phi every d frames;
making a difference with the phi value of the previous d frame;
when the difference value is larger than a preset threshold value T, judging that an abnormality occurs;
if the difference is smaller than the threshold T, repeating the process until the group tracking is finished, and if the difference is still smaller than the threshold T until the group tracking is finished, judging that the group has no abnormal behavior.
4. The system of claim 3, wherein the function φ can be used to perform an average calculation or a variance calculation on the group cooperativity values of the groups within d frames.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510160419.8A CN106157325B (en) | 2015-04-07 | 2015-04-07 | Group abnormal behavior detection method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510160419.8A CN106157325B (en) | 2015-04-07 | 2015-04-07 | Group abnormal behavior detection method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106157325A CN106157325A (en) | 2016-11-23 |
| CN106157325B true CN106157325B (en) | 2020-10-27 |
Family
ID=57337853
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510160419.8A Active CN106157325B (en) | 2015-04-07 | 2015-04-07 | Group abnormal behavior detection method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106157325B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106355605B (en) * | 2016-08-25 | 2018-12-28 | 深圳先进技术研究院 | Group movement consistency filter method |
| CN111401220B (en) * | 2020-03-12 | 2021-03-19 | 重庆特斯联智慧科技股份有限公司 | Crowd aggregation characteristic analysis method and system for intelligent security |
Family Cites Families (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7386420B2 (en) * | 2006-06-02 | 2008-06-10 | United Microelectronics Corp. | Data analysis method for integrated circuit process and semiconductor process |
| CN101872418B (en) * | 2010-05-28 | 2012-09-12 | 电子科技大学 | Detection method based on group environment abnormal behavior |
| CN102799863B (en) * | 2012-07-02 | 2014-12-17 | 中国计量学院 | Method for detecting group crowd abnormal behaviors in video monitoring |
| CN103258193B (en) * | 2013-05-21 | 2016-06-22 | 西南科技大学 | A kind of group abnormality Activity recognition method based on KOD energy feature |
| CN103413321A (en) * | 2013-07-16 | 2013-11-27 | 南京师范大学 | Crowd behavior model analysis and abnormal behavior detection method under geographical environment |
| CN103473533B (en) * | 2013-09-10 | 2017-03-15 | 上海大学 | Moving Objects in Video Sequences abnormal behaviour automatic testing method |
| CN103593646A (en) * | 2013-10-16 | 2014-02-19 | 中国计量学院 | Dense crowd abnormal behavior detection method based on micro-behavior analysis |
| CN103745230B (en) * | 2014-01-14 | 2017-05-10 | 四川大学 | Adaptive abnormal crowd behavior analysis method |
| CN103996051A (en) * | 2014-05-12 | 2014-08-20 | 上海大学 | Method for automatically detecting abnormal behaviors of video moving object based on change of movement features |
| CN104123544B (en) * | 2014-07-23 | 2018-03-13 | 通号通信信息集团有限公司 | Anomaly detection method and system based on video analysis |
-
2015
- 2015-04-07 CN CN201510160419.8A patent/CN106157325B/en active Active
Non-Patent Citations (1)
| Title |
|---|
| Visual abnormal behavior detection based on trajectory sparse reconstruction analysis;CeLi 等;《Neurocomputing》;20131231;第94-100页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106157325A (en) | 2016-11-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US20190378022A1 (en) | Missing value imputation technique to facilitate prognostic analysis of time-series sensor data | |
| CN110689054B (en) | Worker violation monitoring method | |
| KR102217253B1 (en) | Apparatus and method for analyzing behavior pattern | |
| US20100063949A1 (en) | Long-term memory in a video analysis system | |
| CN107346415A (en) | Method of video image processing, device and monitoring device | |
| JP4641450B2 (en) | Unsteady image detection method, unsteady image detection device, and unsteady image detection program | |
| CN104240235A (en) | Method and system for detecting whether camera is covered or not | |
| CN110309735A (en) | Exception detecting method, device, server and storage medium | |
| CN106156706B (en) | Pedestrian abnormal behavior detection method | |
| CN105471659A (en) | Root fault cause analysis method and analysis device | |
| CN103678299A (en) | Method and device for monitoring video abstract | |
| CN101668012B (en) | Method and device for detecting security event | |
| CN112565676A (en) | Method, system and equipment for identifying violation of on-duty personnel based on intelligent camera | |
| CN105338294A (en) | Monitoring device and method | |
| CN106157326B (en) | Group abnormal behavior detection method and system | |
| CN106157325B (en) | Group abnormal behavior detection method and system | |
| CN103107912A (en) | Abnormal detection system based on outlier excavation | |
| CN112818766A (en) | High-altitude parabolic detection alarm method and system based on computer vision | |
| CN112784738A (en) | Moving object detection alarm method, device and computer readable storage medium | |
| Atzori et al. | HEIMDALL: an AI-based infrastructure for traffic monitoring and anomalies detection | |
| US20120257052A1 (en) | Electronic device and method for detecting abnormities of image capturing device | |
| CN111368620B (en) | Method and device for identifying target attention object and storage medium | |
| CN117093461A (en) | Method, system, equipment and storage medium for time delay detection and analysis | |
| CN112581489A (en) | Video compression method, device and storage medium | |
| CN103905789B (en) | The method and its device of the preservation, search and ex-post analysis of dynamic analyze data |
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 |