The Implementation Of Remote Video Monitoring System Of Moving Object Detection Based On Video Sequence

Recently, with the advancement of information, remote video surveillance andmonitoring has become an important research topic, among which target detection andtracking has penetrated into all fields in computer vision. Since the remote videomonitoring system can be widely used in hotel, bank, portable automotive navigationsystem and examination room, remote video surveillance and monitoring has become animportant part of the machine vision. The main purpose of video monitoring system is itcould produce a warning cloud to alert the monitor when moving targets are detected inthe region of interest (ROI) that users have set.Based on the research of currently used target detection algorithm, an improvedViBe algorithm is put forward. On the one hand, an improved ViBe algorithm is putforward on the basis of the conventional target detection algorithm. On the other hand, ageneral remote monitoring system which is based on analog cameras and digitalIPcameras as the prerequisite of static cameras is realized utilizing MFC, C++andOpenCV.The main contents in this thesis are as following:1、Through the study of currently used target detection algorithm and according tothe actual monitoring system and actual requirements, the improved ViBe backgroundextraction algorithm is put forward and the conclusion is given.The experiments showthat not only does the proposed algorithm improve the calculation speed, reduce theimpace of the illumination changing, suppress the target shadow but it can extract thewhole outline of the target and reduce the mistake examining rate, which is suitable forthe practical application of the real-time target detection requirements.2、On the basis of C/S model, a general system has been programmed to possesscomplete functions of the whole system. Analog camera and digital IP camera areapplicable. The main functions of the system being realized are system encryption,video display, video pause, vide zoom, video skip, setting the monitoring zone usersbeing interested in, manual recording and automatic recording, video playback, targetdetection, alarm, YUNTAI control module, snapshots, user permissions setting, the selection of camera type, full screen interface, the minimizing of the interface to the trayand so on.3、The target ViBe detection algorithm is embedded into the system and we havetested single objective results and multi objective results in different monitoringscenarios.The experiments indicate that the system has good generality and has highstability、security relative to the traditional monitoring system. The proposed algorithmcan detect small target under low contrast, improve the detection speed, the accuracy ofthe detection is high, which can meet the need of real-time detection.