Research On Moving Object Detect And Tracking Based On Video Images

Moving objects detecting and tracking based on video images is an important topic of many science domains, including computer vision, computer science and video surveillance, which extends the analysis from single static image to many moving images. Moving objects detecting and tracking technology not only has a promising broad future in human daily work, but also plays a more and more important role in many other domains, such as military, traffic, medicine and scientific research etc. The accuracy and stability of detecting and tracking effects depend on the design of the algorithms. It's still a hot topic to research an robust and accurate detecting and tracking algorithm in computer vision and image compress domains etc.Firstly, this paper introduces some common methods of moving objects detecting and tracking domain, as well as the theoretical basis of digital image processing and moving object detecting and tracking technology. In image preprocessing, this paper introduces image obtain, conversion and noise suppression which lay the solid foundation for moving objects detecting and tracking.In moving objects detecting aspect, according to the character of video image sequences, which has such strong correlation of adjacent frames, this paper uses frame difference algorithm and background difference algorithm. In order to adequately consider the relativity of time and satirically of frequency image sequences,this paper provides a moving objects detection algorithm colligated spatial temporal information. First using background difference algorithm to detect moving objects, and then using symmetry subtractions to detect the moving objects in foreground regions, considering neighborhood information can avoid detecting the background as moving objects, we use background difference based on neighborhood to solve this problem. The result of experiment proves high detection speed and simply operation of the detecting algorithm.In moving objects tracking aspect,based on analysis of moving targets tracking algorithms, this paper discusses and studies Hausdorff measure algorithm which is a staple image matching tracking algorithm, and proposed bidirectional HD measure based on multi-resolution tracking algorithm, the result of experiment proves the accurate of this new tracking algorithm and can solve the compound targets covered problem, which has certain robustness.