CN210072642U - Crowd abnormal behavior detection system based on video monitoring - Google Patents
Crowd abnormal behavior detection system based on video monitoring Download PDFInfo
- Publication number
- CN210072642U CN210072642U CN201920941394.9U CN201920941394U CN210072642U CN 210072642 U CN210072642 U CN 210072642U CN 201920941394 U CN201920941394 U CN 201920941394U CN 210072642 U CN210072642 U CN 210072642U
- Authority
- CN
- China
- Prior art keywords
- early warning
- crowd
- abnormal
- video
- characteristic
- 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
Abstract
The crowd abnormal behavior detection system based on video monitoring comprises a plurality of acquisition cameras which are connected in parallel; the abnormality detection device carries out abnormality detection on the video sequence image acquired by the video acquisition device; the early warning device carries out early warning according to the abnormity detection result of the abnormity detection device. The anomaly detection device comprises a feature extractor, a feature encoder and an anomaly detector, wherein the feature extractor is used for extracting the appearance features of the crowd according to the video sequence image acquired by the video acquisition device; the characteristic encoder is connected with the characteristic extractor and encodes the extracted crowd appearance characteristic image through a perceptual hash algorithm to obtain a characteristic fingerprint; the anomaly detector is connected with the characteristic encoder, and determines the anomaly degree of the crowd appearance characteristic image according to the similarity of the characteristic fingerprints obtained by the characteristic encoder. According to the technical scheme, the abnormal behaviors of the crowd can be found in time and early-warning is carried out, the public safety of the society is guaranteed, and the harm caused by the abnormal behaviors of the crowd is avoided.
Description
Technical Field
The embodiment of the utility model provides a relate to security protection control technical field, concretely relates to crowd abnormal behavior detecting system based on video monitoring.
Background
At present, intelligent video monitoring becomes one of the very active hot techniques in the field of artificial intelligence, is an important application of the computer vision technology in the field of security protection, and realizes the conversion from the 'passive monitoring' to the 'active monitoring' of the traditional video monitoring. China's economy develops rapidly, and town population gathers constantly, has often had the people stream peak in public areas such as railway station, subway station, square, however the crowd of higher density has brought very big hidden danger and pressure to public safety, and the urgency nature of safety precaution, safety monitoring and safety protection work is aggravated constantly.
In recent years, group abnormal behaviors such as group fighting and illegal aggregation occur, and great threat is brought to public safety of the society. Therefore, group abnormal behavior detection and online early warning for a monitored scene become important leading-edge subjects in the fields of computer vision, pattern recognition and the like. Furthermore, due to the complexity of crowd motion and the diversity of abnormal events in videos, it is difficult to accurately describe abnormal behavior in videos. Secondly, video monitoring is mostly deployed in public places with dense crowds, monitoring scenes are various, and individuals in the crowds are shielded from each other, so that extracted behavior characteristics are inaccurate, and abnormal behavior modeling is difficult. There is a need for a system for detecting abnormal behaviors of people based on video monitoring.
SUMMERY OF THE UTILITY MODEL
Therefore, the embodiment of the utility model provides a crowd abnormal behavior detecting system based on video monitoring can acquire public place crowd's image and carry out the abnormal detection early warning, guarantees social public safety.
In order to achieve the above object, the embodiment of the present invention provides the following technical solutions: the crowd abnormal behavior detection system based on video monitoring comprises a video acquisition device, an abnormal detection device, a communication device and an early warning device; the video acquisition device is connected with the abnormality detection device and comprises a plurality of acquisition cameras which are connected in parallel based on a TCP/IP protocol; the abnormality detection device is connected with the communication device and is used for carrying out abnormality detection on the video sequence image acquired by the video acquisition device; the early warning device is connected with the communication device and used for early warning according to the abnormity detection result of the abnormity detection device.
As a preferred scheme of the crowd abnormal behavior detection system based on video monitoring, the video acquisition device adopts a third generation video monitoring system IPVS, and the acquisition camera is internally provided with a Web server and an Ethernet port.
As a preferred scheme of a crowd abnormal behavior detection system based on video monitoring, the abnormal detection device comprises a feature extractor, a feature encoder and an abnormal detector, wherein the feature extractor is used for extracting the crowd appearance features according to a video sequence image acquired by the video acquisition device; the characteristic encoder is connected with the characteristic extractor and is used for encoding the extracted crowd appearance characteristic image through a perceptual hash algorithm to obtain a characteristic fingerprint; the anomaly detector is connected with the characteristic encoder and used for determining the anomaly degree of the crowd appearance characteristic image according to the similarity of the characteristic fingerprints obtained by the characteristic encoder.
As a preferable scheme of the crowd abnormal behavior detection system based on video monitoring, the feature extractor extracts local optical flow features of the target video sequence image based on a Shi-Tomasi feature point enhanced Lucas-Kanade optical flow method.
As a preferable scheme of the crowd abnormal behavior detection system based on video monitoring, the abnormal detector is used as a crowd appearance characteristic image abnormal behavior judgment basis through inter-frame similarity and optical flow value fusion.
As a preferred scheme of a crowd abnormal behavior detection system based on video monitoring, the early warning device comprises a fixed early warning terminal, and the fixed early warning terminal carries out early warning according to an abnormal detection result of the abnormal detection device;
the fixed early warning terminal is connected with a broadcast dispersion device, and the broadcast dispersion device is used for broadcasting dispersion to people according to the abnormal detection result of the abnormal detection device.
As a preferred scheme of a crowd abnormal behavior detection system based on video monitoring, the early warning device comprises a mobile early warning terminal, and the mobile early warning terminal carries out early warning according to an abnormal detection result of the abnormal detection device;
the mobile early warning terminal is connected with a broadcast dispersion device, and the broadcast dispersion device is used for carrying out broadcast dispersion on crowds according to the abnormal detection result of the abnormal detection device.
As a preferable scheme of the crowd abnormal behavior detection system based on video monitoring, the communication device adopts a 3G, 4G or 5G mobile communication network.
The embodiment of the utility model provides a have following advantage: the system is provided with a video acquisition device, an abnormality detection device, a communication device and an early warning device; the video acquisition device is connected with the abnormality detection device and comprises a plurality of acquisition cameras which are connected in parallel based on a TCP/IP protocol; the abnormality detection device is connected with the communication device and is used for carrying out abnormality detection on the video sequence image acquired by the video acquisition device; the early warning device is connected with the communication device and used for early warning according to the abnormity detection result of the abnormity detection device. The anomaly detection device comprises a feature extractor, a feature encoder and an anomaly detector, wherein the feature extractor is used for extracting the appearance features of the crowd according to the video sequence image acquired by the video acquisition device; the characteristic encoder is connected with the characteristic extractor and used for encoding the extracted crowd appearance characteristic image through a perceptual hash algorithm to obtain a characteristic fingerprint; and the anomaly detector is connected with the characteristic encoder and is used for determining the anomaly degree of the crowd appearance characteristic image according to the similarity of the characteristic fingerprints obtained by the characteristic encoder. By the technical scheme, abnormal behaviors of the crowd can be found in time and early warning is carried out, social public safety is guaranteed, and harm caused by abnormal crowd is avoided.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below. It should be apparent that the drawings in the following description are merely exemplary, and that other embodiments can be derived from the drawings provided by those of ordinary skill in the art without inventive effort.
Fig. 1 is a schematic view of a system for detecting abnormal behaviors of people based on video monitoring provided in an embodiment of the present invention;
fig. 2 is a schematic diagram of a feature encoder provided in an embodiment of the present invention;
in the figure: 1. a video capture device; 2. an abnormality detection device; 3. a communication device; 4. an early warning device; 5. Collecting a camera; 6. fixing the early warning terminal; 7. moving the early warning terminal; 8. a feature extractor; 9. a feature encoder; 10. an abnormality detector; 11. broadcast grooming device.
Detailed Description
The present invention is described in terms of specific embodiments, and other advantages and benefits of the present invention will become apparent to those skilled in the art from the following disclosure. Based on the embodiments in the present invention, all other embodiments obtained by a person skilled in the art without creative work belong to the protection scope of the present invention.
It should be further noted that, in the technical solution, what is claimed is to implement hardware deployment of a crowd abnormal behavior detection system based on video monitoring, and protection is not required to be performed on related algorithms or methods, and in addition, the related algorithms or methods will be fully disclosed as prior art in the embodiments, so that those skilled in the art can better implement the technical solution. In analogy, the algorithm or method related to the technical solution is equivalent to a workflow and a using method, and any structural technical solution necessarily has a corresponding implementation method in the implementation or implementation process, and on the premise that the applicant does not request protection of the method, the structural technical solution cannot be denied as an object for protection of the utility model due to the related method.
Referring to fig. 1, the crowd abnormal behavior detection system based on video monitoring comprises a video acquisition device 1, an abnormal detection device 2, a communication device 3 and an early warning device 4, wherein the communication device 3 adopts a 3G, 4G or 5G mobile communication network; the video acquisition device 1 is connected with the abnormality detection device 2, and the video acquisition device 1 comprises a plurality of acquisition cameras 5 which are connected in parallel based on a TCP/IP protocol; the abnormality detection device 2 is connected with the communication device 3, and the abnormality detection device 2 is used for performing abnormality detection on the video sequence image acquired by the video acquisition device 1; the early warning device 4 is connected with the communication device 3, and the early warning device 4 is used for early warning according to the abnormity detection result of the abnormity detection device 2.
In an embodiment of the system for detecting the abnormal behavior of the crowd based on video monitoring, the video acquisition device 1 adopts a third generation video monitoring system IPVS, and the acquisition camera 5 is internally provided with a Web server and an ethernet port. The acquisition camera 5 generates JPEG or MPEG4, h.264 data files for access, monitoring, recording and printing by authorized terminals from any location in the network. The acquisition camera 5 is simply connected to the network through a wired or wireless ethernet, and can transmit the output image of the camera and horizontal, vertical and zoom (PTZ) control commands by using five types of network cables or wireless network modes.
In an embodiment of the system for detecting the abnormal behavior of the crowd based on video monitoring, the early warning device 4 includes a fixed early warning terminal 6, and the fixed early warning terminal 6 performs early warning according to an abnormal detection result of the abnormal detection device 2. The early warning device 4 comprises a mobile early warning terminal 7, and the mobile early warning terminal 7 carries out early warning according to the abnormity detection result of the abnormity detection device 2. The fixed early warning terminal 6 and the mobile early warning terminal 7 are connected with a broadcast persuasion device 11, and the broadcast persuasion device 11 is used for broadcasting to crowd according to the abnormal detection result of the abnormal detection device 2, so that the crowd is effectively dredged, and the safety of the crowd is ensured.
Referring to fig. 2, in an embodiment of the system for detecting the abnormal behavior of the crowd based on video monitoring, the abnormal detection device 2 includes a feature extractor 8, a feature encoder 9 and an abnormal detector 10, where the feature extractor 8 is configured to perform crowd appearance feature extraction according to a video sequence image acquired by the video acquisition device 1; the feature encoder 9 is connected with the feature extractor 8, and the feature encoder 9 is used for encoding the extracted crowd appearance feature image through a perceptual hash algorithm to obtain a feature fingerprint; the anomaly detector 10 is connected with the feature encoder 9, and the anomaly detector 10 is used for determining the anomaly degree of the crowd appearance feature image according to the similarity of the feature fingerprints obtained by encoding of the feature encoder 9.
Specifically, the perceptual hash algorithm is a generic term of a class of algorithms, and can be used for searching similar pictures. The perceptual hash algorithm may generate a "fingerprint" (fingerprint) string for each image, and compare fingerprint information of different images to determine similarity of the images. The closer the result is to the image the more similar. Perceptual hash algorithms include mean hash (aHash), perceptual hash (pHash), and dHash (differential hash). aHash speed is faster, but accuracy is lower; pHash is performed in reverse, with higher accuracy but slower speed; the dHash takes both into account, and has higher accuracy and higher speed. After a 64-bit hash value is obtained, the hamming distance is used to quantify the similarity of the two images. The greater the hamming distance is, the smaller the similarity of the images is, and the smaller the hamming distance is, the greater the similarity of the images is.
In an embodiment of the system for detecting the abnormal behavior of the crowd based on video monitoring, the feature extractor 8 extracts the local optical flow features of the target video sequence image based on a Shi-Tomasi feature point enhanced Lucas-Kanade optical flow method. Specifically, in the aspect of extracting the appearance features of the video, the technical scheme extracts the local motion features by using a pyramid Lucas-Kanade optical flow method. In addition, in order to improve the capability of characterizing the salient motion features, an LOF (local Outlier factor) anomaly detection algorithm is applied to the detection of the corner features to increase the number of abnormal corners. The corner point detection method uses Shi-Tomasi. The feature extractor 8 extracts features from two aspects of appearance similarity and local motion light flow from the aspect of video multi-feature extraction, and finally models the fusion abnormal value through a single classification SVM model so as to realize the effect of abnormal detection. The feature encoder 9 extracts high-level semantic features of the images by using a deep convolutional network AlexNet, extracts feature map fingerprints by using a perceptual hash algorithm, thereby obtaining appearance feature fingerprints, quantizes the similarity between the images according to the similarity of the fingerprints, effectively reduces the computational complexity, and obtains a good detection effect. The anomaly detector 10 is used as a basis for judging the anomaly behavior of the crowd appearance feature image through the fusion of the inter-frame similarity and the optical flow value.
Specifically, the Shi-Tomasi algorithm is an improvement of the Harris algorithm. The Harris algorithm is most originally defined by subtracting the determinant value of the matrix M from the trace of M and comparing the difference with a predetermined threshold. Later Shi and Tomasi proposed improved methods that would result in strong corners if the smaller of the two eigenvalues was greater than a minimum threshold. And carrying out eigenvalue analysis on the autocorrelation matrix M to generate two eigenvalues and two eigenvector vectors. Since a larger uncertainty depends on a smaller eigenvalue, better results than using the Harris algorithm can be obtained by finding the maximum of the smallest eigenvalue to find a good eigenvalue. The Lucas-Kanade optical flow algorithm is a two-frame differential optical flow estimation algorithm proposed by BruceD. The optical flow algorithm can evaluate the deformation between two images, the basic assumption is that voxels and image pixels are conserved, and the image constraint equation can be obtained based on the idea that the color of an object does not change greatly and obviously in two frames. Different optical flow algorithms solve the optical flow problem assuming different additional conditions. It is understood by those skilled in the art that Optical flow (Optical flow) is a motion pattern that refers to the apparent movement of an object, surface, edge between an observer (e.g., eye, camera, etc.) and the background at a viewing angle.
The embodiment of the utility model provides a video acquisition device 1, unusual detection device 2, communication device 3 and early warning device 4 are equipped with; the video acquisition device 1 is connected with the abnormality detection device 2, and the video acquisition device 1 comprises a plurality of acquisition cameras 5 which are connected in parallel based on a TCP/IP protocol; the abnormality detection device 2 is connected with the communication device 3, and the abnormality detection device 2 is used for carrying out abnormality detection on the video sequence image acquired by the video acquisition device 1; the early warning device 4 is connected to the communication device 3, and the early warning device 4 is used for giving an early warning according to the abnormality detection result of the abnormality detection device 2. The anomaly detection device 2 comprises a feature extractor 8, a feature encoder 9 and an anomaly detector 10, wherein the feature extractor 8 is used for extracting the appearance features of the crowd according to the video sequence image acquired by the video acquisition device 1; the feature encoder 9 is connected with the feature extractor 8, and the feature encoder 9 is used for encoding the extracted crowd appearance feature image through a perceptual hash algorithm to obtain a feature fingerprint; the anomaly detector 10 is connected with the feature encoder 9, and the anomaly detector 10 is used for determining the anomaly degree of the crowd appearance feature image according to the similarity of the feature fingerprints obtained by encoding of the feature encoder 9. Compared with the prior art, the anomaly detection device 2 in the technical scheme measures the similarity from two aspects of appearance similarity and local motion optical flow respectively from the aspects of video processing full flow, multi-feature extraction and anomaly capture algorithm; in the aspect of appearance similarity calculation, a perceptual hash algorithm is used for measuring the inter-frame similarity, so that the video similarity comparison efficiency is greatly improved; during motion feature calculation, extracting local optical flow features of the target by using an improved Lucas-Kanade optical flow method based on Shi-Tomasi feature point enhancement; in the aspect of abnormal behavior detection, a single classification SVM model is used for making abnormal value decision, so that the influence of uncertain factors such as threshold setting on abnormal detection is avoided; in the aspect of early warning result feedback, a B/S framework online feedback and short message reminding mode is adopted, and the man-machine interaction performance is more flexible. The technical scheme is more comprehensive in the aspect of specific technical means, and has better technical effect than other similar technologies. By the technical scheme, abnormal behaviors of the crowd can be found in time and early warning is carried out, social public safety is guaranteed, and harm caused by abnormal crowd is avoided.
Although the invention has been described in detail with respect to the general description and the specific embodiments, it will be apparent to those skilled in the art that modifications and improvements can be made based on the invention. Therefore, such modifications and improvements are intended to be within the scope of the invention as claimed.
Claims (5)
1. The crowd abnormal behavior detection system based on video monitoring is characterized by comprising a video acquisition device (1), an abnormal detection device (2), a communication device (3) and an early warning device (4); the video acquisition device (1) is connected with the abnormality detection device (2), and the video acquisition device (1) comprises a plurality of acquisition cameras (5) which are connected in parallel based on a TCP/IP protocol; the abnormality detection device (2) is connected with the communication device (3), and the abnormality detection device (2) is used for performing abnormality detection on the video sequence image acquired by the video acquisition device (1); the early warning device (4) is connected with the communication device (3), and the early warning device (4) is used for carrying out early warning according to the abnormity detection result of the abnormity detection device (2).
2. The system for detecting the crowd abnormal behavior based on the video monitoring as claimed in claim 1, wherein the video capturing device (1) adopts a third generation video monitoring system IPVS, and the capturing camera (5) is provided with a Web server inside and an Ethernet port.
3. The video monitoring based crowd abnormal behavior detection system according to claim 1, wherein the early warning device (4) comprises a fixed early warning terminal (6), and the fixed early warning terminal (6) performs early warning according to an abnormal detection result of the abnormal detection device (2);
the fixed early warning terminal (6) is connected with a broadcast dispersion device (11), and the broadcast dispersion device (11) is used for broadcasting dispersion to people according to the abnormal detection result of the abnormal detection device (2).
4. The video monitoring based crowd abnormal behavior detection system according to claim 1, wherein the early warning device (4) comprises a mobile early warning terminal (7), and the mobile early warning terminal (7) performs early warning according to an abnormal detection result of the abnormal detection device (2);
the mobile early warning terminal (7) is connected with a broadcast dispersion device (11), and the broadcast dispersion device (11) is used for broadcasting dispersion to people according to the abnormal detection result of the abnormal detection device (2).
5. The video surveillance-based crowd abnormal behavior detection system according to claim 1, wherein the communication device (3) employs a 3G, 4G or 5G mobile communication network.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201920941394.9U CN210072642U (en) | 2019-06-20 | 2019-06-20 | Crowd abnormal behavior detection system based on video monitoring |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201920941394.9U CN210072642U (en) | 2019-06-20 | 2019-06-20 | Crowd abnormal behavior detection system based on video monitoring |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN210072642U true CN210072642U (en) | 2020-02-14 |
Family
ID=69428480
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201920941394.9U Expired - Fee Related CN210072642U (en) | 2019-06-20 | 2019-06-20 | Crowd abnormal behavior detection system based on video monitoring |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN210072642U (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113382204A (en) * | 2021-05-22 | 2021-09-10 | 特斯联科技集团有限公司 | Intelligent processing method and device for fire-fighting hidden danger |
| CN115484456A (en) * | 2022-09-15 | 2022-12-16 | 重庆邮电大学 | Video anomaly prediction method and device based on semantic clustering |
-
2019
- 2019-06-20 CN CN201920941394.9U patent/CN210072642U/en not_active Expired - Fee Related
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113382204A (en) * | 2021-05-22 | 2021-09-10 | 特斯联科技集团有限公司 | Intelligent processing method and device for fire-fighting hidden danger |
| CN115484456A (en) * | 2022-09-15 | 2022-12-16 | 重庆邮电大学 | Video anomaly prediction method and device based on semantic clustering |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US20200364443A1 (en) | Method for acquiring motion track and device thereof, storage medium, and terminal | |
| CN110942545B (en) | Dense person entrance guard control system and method based on face recognition and video fence | |
| CN103824070B (en) | A kind of rapid pedestrian detection method based on computer vision | |
| CN111832457B (en) | Stranger intrusion detection method based on cloud edge cooperation | |
| CN108564052A (en) | Multi-cam dynamic human face recognition system based on MTCNN and method | |
| CN105205455A (en) | Liveness detection method and system for face recognition on mobile platform | |
| Yu et al. | Motion pattern interpretation and detection for tracking moving vehicles in airborne video | |
| WO2019114145A1 (en) | Head count detection method and device in surveillance video | |
| CN108416256A (en) | The family's cloud intelligent monitor system and monitoring method of feature based identification | |
| Cao et al. | Abnormal crowd motion analysis | |
| KR20150112713A (en) | The predicting system for anti-crime through analyzing server of images | |
| Karpagavalli et al. | Estimating the density of the people and counting the number of people in a crowd environment for human safety | |
| CN210072642U (en) | Crowd abnormal behavior detection system based on video monitoring | |
| CN109948474A (en) | AI thermal imaging all-weather intelligent monitoring method | |
| CN112287823A (en) | Facial mask identification method based on video monitoring | |
| Wang et al. | Early smoke detection in video using swaying and diffusion feature | |
| Zaidi et al. | Video anomaly detection and classification for human activity recognition | |
| Chen et al. | A dataset and benchmark towards multi-modal face anti-spoofing under surveillance scenarios | |
| Dahmane et al. | Real-time video surveillance with self-organizing maps | |
| CN113537050A (en) | Dynamic face recognition algorithm based on local image enhancement | |
| KR101485512B1 (en) | The sequence processing method of images through hippocampal neual network learning of behavior patterns in case of future crimes | |
| Chen et al. | An intelligent virtual fence security system for the detection of people invading | |
| Krishna et al. | Advanced video surveillance system | |
| Kroneman et al. | Accurate pedestrian localization in overhead depth images via Height-Augmented HOG | |
| Hsieh et al. | Grid-based template matching for people counting |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20200923 Address after: 830000 No. 666 Shengli Road, Urumqi City, Xinjiang Uygur Autonomous Region Patentee after: XINJIANG University Address before: 830046 No. 666 Shengli Road, Tianshan District, the Xinjiang Uygur Autonomous Region, Urumqi Patentee before: Ma Guoxiang |
|
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20200214 Termination date: 20210620 |