100% found this document useful (1 vote)
275 views

Research On Bank Intelligent Video Image Processing and Monitoring Control System Based On OpenCV

This intelligent video monitoring system for banks uses OpenCV for real-time image processing and analysis. The system detects humans using background subtraction and tracks detected motions. It analyzes behaviors using spatial and temporal characteristics of tracked body positions to identify behaviors. This provides intelligent monitoring and alarm within the bank, improving security over traditional video systems.

Uploaded by

thangaraj_ic
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
100% found this document useful (1 vote)
275 views

Research On Bank Intelligent Video Image Processing and Monitoring Control System Based On OpenCV

This intelligent video monitoring system for banks uses OpenCV for real-time image processing and analysis. The system detects humans using background subtraction and tracks detected motions. It analyzes behaviors using spatial and temporal characteristics of tracked body positions to identify behaviors. This provides intelligent monitoring and alarm within the bank, improving security over traditional video systems.

Uploaded by

thangaraj_ic
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 4

Research on bank intelligent video image processing and monitoring control system based on OpenCV

Yongxiang Wu
college of mechanical and electrical engineering Guizhou Normal University Guiyang,China [email protected]
AbstractAs the principal place for currency flow nowadays, the banking system plays a special role in society and its security has always been a focus. The current bank branches security systems usually use video surveillance system, which is basically used to provide evidence after incidents, and it can not do realtime warning. With the rapid development of computer vision, video and intelligent recognition, real-time alarm analysis of surveillance system has become possible. Based on the technologies of human body detection, motion tracking and behavior judgement, this article designs a bank intelligent video image processing and monitoring control system which is based on OpenCV ,and the system not only realizes real-time monitoring and alarm within monitoring zone effectively, but also brings innovations to the traditional video surveillance system, providing strong protection of safe operation for banks. Keywords-OpenCV; human detection; motion tracking; behavior judgement; intelligent monitoring

realize the common image processing and computer vision algorithms. OpenCV is compatible with Intel Image Processing Library (IPL) which is another image processing library and is also developed by the Intel, and IPL is used to realize some digital image processing at low level, while OpenCV is mainly used in some advanced image processing [5] ,including feature detection and tracking, motion analysis, object segmentation and recognition ,and 3D reconstruction etc. As the source codes of OpenCV are shared completely and are both simple and efficient, together with compilation optimized functions, it has been widely used in image processing fields in foreign countries these years. Compared with existing computer vision libraries, OpenCV has the following advantages: OpenCV is an application interface containing more than 300 C programming functions, and it does not depend on external libraries, which means it can run not only independently but also with other external libraries. All of the algorithms in OpenCV are based on the dynamic data structures packed in the IPL, which have high flexibility and more than half of them have been optimized for Intel processors when they are designed and complied, which makes them high efficient. Providing interfaces with some other language or platforms such as EiC, Ch, MATLAB, all of the interfaces will be installed under the directory of OpenCV. The source codes of OpenCV are open, and developers can modify the codes or add new classes into the library.The codes can also be used by other developers as long as they meet the design standards. All of these features have given OpenCV great vitality and development potential. III. CONSTRUCTION OF INTELLIGENT VIDEO IMAGE PROCESSING AND MONITORING CONTROL SYSTEM

I.

INTRODUCTION

In recent years, with the increasing security protection demands of bank branches, video surveillance [1] system has been widely used as an important component of the security system. However, the existing video surveillance systems in banks are mostly used for images evidence collection after incident, and it can not provide intelligent real-time image processing for early warning. With the rapid development of computer vision and video intelligent identification, its now possible to integrate human body recognition technology with video surveillance system to achieve intelligent video surveillance [2][3]. Therefore, the paper designed an intelligent bank video processing surveillance system based on the OpenCV development platform. With the integration of computer vision and pattern recognition, the system can realize the analysis and report of abnormal behaviors in banks, thus it improves security of monitoring region greatly and provides strong protection for banks safe operation. II. BRIEF INTRODUCTION OF OPENCV

The open source Computer Vision Library OpenCV (Intel Open Source Computer Vision Library) has been developed by the Intel research lab in Russia [4], it consists of several libraries of C functions and C + + classes, and it is used to

This intelligent monitoring system screen image processing is composed of both hardware and software. The hardware part includes camera, video capture card and computer. The camera is responsible for image acquisition and transmission to the

video capture card, then the video capture cards would apply ADC process to the images and send the results to the computer where body motion information is obtained through computer-related algorithms and analyzed by software. The software consists of behavior detection and motion tracking algorithm which are the central parts in the entire system. The video image sent by the camera would go through preprocessing, processing, post-processing and analysis before final disposal results come out. Thus, it can be seen that the realization of the video processing algorithms is the key point of system construction. This part has been realized by combination of the OpenCV development platform and algorithms of human body detection, motion tracking and behavior determination. The flow chart of the entire system is shown in Figure 1.

A. Human motion detection The critical part of human motion detection is the description model of background image or the background model, which is base of pre-image object division in the background subtraction algorithm. The background model has single-mode and multi-modal [6], the former can be described with a single probability distribution model as the color distribution of each background spots is relatively more compact; the distribution of the latter is more dispersed, so multiple distribution models are needed. The human body detection in this system mainly makes use of background updating algorithm to obtain background image, and then subtracts it with current frame to get the region of human body which would be made to binary image and get filtered, and finally use the two-valued image segmentation technology to locate of human body region. The background update algorithm is mainly used for image extraction and updating here. By calling relevant OpenCV functions, we can do modeling and detection to the background, the realization codes are: // take video If(!(pCapture=CaptureFromFile(VideoStream))) {fprintf(stderr, can not open the video stream) Return -2} // read video frame-by-frame While(pFrame = sscvQueryFrame(pCaptuer)) { nFrmNum++; // Gaussian filtering scvSmooth(pFrameMat,pFrameMat,SCV_GAUSSINA,3,0,0); // subtract background with current frame scvAbsDiff(pFrameMat,pBkMat,pFrMat); // binarization foreground scvThreshold(pFrmat,pFrImg,60,255,0,SCV_THRESH_BINA RY); //noise elimination scvErode(pFrImg,pFrImg,0,1); scvDilate(pFrImg,pFrImg,0,1); //background update and conversion, detection of Human Motion scvRunningAvg(pFrameMat,pBkMat,0,003,0); scvConVert(pBkMat,pBkImg);

Figure 1. Flow chart of the entire system

//display scvShowImage(video,pFrame); scvShowImage(background,pBkImg);

scvShowImage(foreground,pFrImg); } In order to examine the effects of image extraction and updating, we extracted and updated the image of same place in the morning and afternoon respectively.The results are shown in Figure 2 and Figure 3. It can be seen that the expected image has been well obtained.

The realization of human body tracking and behavior detection algorithm is: the body region is extracted from each frame image, after that a rectangle is used to represent the region, and the center of the rectangle will serve as the tracking feature point. According to the position variation of center point, track of the motion can be calculated and be drawn as curves. Then the ratio of horizontal and vertical projections is obtained to identify body gestures and describe motion behaviors. The structure is shown in Figure 4. We adopted background method in foreground detection for the objects considered in this system [10] which possesses robustness against noise. According to operation results, detection is well achieved, as shown in Figure 5. It can also been seen that the background method based on OpenCV not only inherits but also improves its robustness against noise.It is shown that only one noise point appears in the second map.

Figure 2. Test result in the morning

Figure 3. Test result in the afternoon

B. Motion tracking and behavior to determine Automatic behavior understanding and analysis is the key of intelligent video monitoring realization. There are mainly three methods of behavior analysis: analysis method based on human body, method based on the temporal and spatial characteristics and method based on image statistical analysis. The method which is based on temporal and spatial characteristics firstly extracts human body zone in each frame, and then try to interpret and indentify the human behavior with various form characteristics of the human outline sequence of the action, for example, Bobick tried to describe human behavior by extracting the Hu constant moment from history motion image [7], while some othersanalyzed human movement by extracting various geometry characteristics of space-time edge line, consisting of all body outlines, from motion sequence [8] [9]. Such methods rely on accurate extraction of the body region in each frame image. The method introduced in this paper is a reform of the temporal and spatial characteristics based method, by combining with OpenCV functions.

Figure 4. chart of structure

monitoring zone, and it works through LAN. At the same time, a signal would also be sent to the local security center, leading to realizing simultaneous alarm. Intelligent video surveillance of banks is realized through this way. IV.
CONCLUSION

Figure 5. the foreground of test results

As for the behavior feature extraction of surveillance video section, the system firstly extracts the directions of mass movement in every pre-image frame ,and then it converts the mass into binary data in accordance with the directions, as shown in Figure 6.

Starting with the technologies of human body detection, motion tracking and behavior judgement and combining computer version with pattern recognition, this article designs a Bank Intelligent Video image processing and Monitoring Control System, based on the OpenCV vision library. The system has a strong adaptability for the varieties in the monitoring region. It is able to ensure that background images which are based on static areas refresh will be updated in time. And by reforming the temporal and spatial characteristics analysis method, it realizes good judgement of pre-image frame,which also profits from the robustness against noise of background method. Thus, accurate detection of unusual actions is guaranteed. The system also fills the blank of realtime alarm monitoring system in banks, greatly improving the safety of surveillance area and providing strong protection for banks. REFERENCES
Collins. R. T, Lipton. A. J, Kanade. T., Introduction to thespecial section on video surveillance, IEEE Transactionson Pattern Analysis and Machine Intelligence, vol. 22, pp. 745746, August 2000 [2] Sudo. K, Osawa. T, Wakabayashi. K, Koike. H., Detecting the degree of anomaly in security video sequences, In: Proceed-ings of Signal and Image Processing. Tokyo, Japan: ACTA Press, pp. 3-7, August 2006 [3] Gao. J, Collins. R. T, Hauptmann. A. G, Wactlar. H. D, Articu-lated motion modeling for activity analysis, In: Proceedings of the 2004 Conference on Computer Vision and Pattern Recognition Workshop. Washington D. C., USA, vol. 1, pp. 2020, July 2004 [4] Song, Li, Xijian. Ping, Yihong. Ding, Application of open source computer vision library, Computer Applications and Software, vol. 8, pp. 134-136, August 2005 [5] Open Source Computer Image Library Reference Mannual.2001(8) [6] Wren. C, Azarbayejani. A, Darrell. T, et. al.. Pfinder; Real-time tracking of the human boby [J], IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 19, pp. 780-785, July 1997 [7] Bobick. A. F, Davis. J. W., The recognition of human move-ment using temporal templates, IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 23, pp. 257267, March 2001 [8] Weinland. D, Ronfard. R, Boyer. E, Free viewpoint action recognition using motion history volumes, Computer Vision and Image Understanding, vol. 104, pp. 249257, November 2006 [9] Yilmaz. A, Shah. M, Actions sketch: a novel action repre-sentation, In: Proceedings of IEEE Computer Society Con-ference on Computer Vision and Pattern Recognition. SanDiego, USA, vol. 1, pp. 984989, June 2005 [10] Stauer. C, Grimson. W. E. L, Adaptive background mixture models for real-time tracking, In: Proceedings of IEEE Com-puter Society Conference on Computer Vision and Pattern Recognition. Ford Collins, USA, vol. 2, pp. 246252, June 1999 [1]

Figure 6. picture of video section mass binary

The behavior features of this video section would be obtained afterwards. The differences of various actions can be interpreted effectively through the binary convert of mass movement directions. The last step is to perform behavior detection of video section by outlining the binary image. Normal and abnormal behaviors can be well distinguished by using the features acquired by such method. Confirmation of unusual actions relies on both whether the security line has been crossed from one side or both sides and whether there is an entrance or a departure in the security zone. All of these procedures can be realized by calling related functions, such as double, cvblobdetector and so on. Finally, once abnormal actions have been observed in video section, real-time warning would be realized by sending the signals to the speaker which is installed on the top of the

You might also like