Study On The Algorithm And Application Of Slow Moving Target Feature Tracking In Natural Scene

Briefly, video moving target tracking technology is tracing the information, such as location, speed and accelerated speed, of the interested target in the next image. It will be the foundation for deeply processing in analyzing tracing trend and moving behavior understanding. Video moving target tracking technology has widely used in many area toady, such as military and medicine. For the military area, this technique can help army searching and locking the enemy’s moving track in a quick and stable way. It also can be used for real-time tracking the targeted histolytic in medicine area.Although the research in moving target-tracking technology has developed for many years, it stills a very hot topic in the field of science.Firstly, this dissertation will discuss the domestic and overseas current research status about the video moving target tracking technology, which including the deep analyzing the hardest part about the today’s research. It analyzes the classical SIFT feature points matching based target tracking algorithm and Mean Shift target tracking algorithm as well. The algorithm base on SIFT feature points will obtain strong robustness in a high image resolution situation, while the robustness will be restricted when the image resolution decreased. The Mean Shift target tracking algorithm can get high real-time performance, but it might be decreased the tracing accuracy over time as the target location information not be used efficiently enough.Then, the dissertation will discuss a video moving target tracking algorithm,which based on the DWT-DCT coefficient feature matching. This algorithm locates the target through feature detection, which will not only use the image information of the target object, but also make most of location information. Comparing the experimental analysis of this algorithm with SIFT feature points matching based target tracking algorithm and the Mean Shift target tracking algorithm, video moving target tracking algorithm can keep strong robustness and can tracing small target accurately even in a low resolution condition.Image super-resolution, which arms to get the desired sequence of images, is the subsequent application of target tracking technology. However, because of the status of current surveillance video, it might not be clear enough for the detail informationof the moving target. So in order to get more clear information of the moving target, it needs to get sequence of images via image super-resolution reconstruction algorithm.In this essay, it presents the universality and effectiveness of the moving target tracing technology in practical application through the super-resolution experiment for the sequences image that obtained by tracking.