Research On The Detection And Tracking Of Moving Target In Video Monitoring System

Video monitor system becomes more and more important in our life,and people think highly ofits use in battlefield surveillance,village public security,road support and so on. In this paper,the mainresearch lays on detecting and tracking dynamic targets in video monitoring.Here are the mainworkings:In the working of fore-manage, the main work laid on the research of wiping noise off anddetecting edge. A fore-judge mean value algorithm is proposed and the performance in wiping salt&papper noise off promoted remarkably. A four direction prewitt method whose basis is the otsuthreshold is proposed, which promote the effect of detecting edge.In the research of detecting dynamic target, a binaryzation whose basis is block threshold isproposed, binary the target area and the non-target area with different threshold, combined the targetoutline which used three frames difference, good detecting results have been achieved.In the research the issue of shadow detecting, A shadow detect method is proposed, consider thesaturation component, luminance component and the two components’ normalized difference todetect shadow, good detecting results have been achieved.In the research of the corner detect, a fast algorithm is proposed. Fore judge all pixels in a imagefirst, wipe off the pixels which couldn’t be corners, use Harris algorithm to detect the rest of candidatecorners. Good results and timeliness have been achieved.In the research of tracking dynamic target. Given classic Mean-Shift used the color feature totrack, it could lose target when the color difference between the target and the background is small.Considered the corners’ characteristics, a combined algorithm was proposed, which improve therobust of tracking target.