An Extraction Method Of Video Objects In Noise Video Sequences

Video coding technology has experienced about thirty years of development. Traditional wave-based coding has been successfully changed into mode-based coding that has been a dominant tendency to video coding technology. The new video coding standard MPEG-4 was finished in 1998, which is based on content coding and aims at improving compression ratio and flexible access to video data. In order to support content-base functions in the visual standard of MPEG-4, each frame of video sequences should be represented in terms of video object plane from the video sequence. So segmentation and tracking of video moving objects become a key research topic in the field of interactive multimedia.This paper focuses on extraction and tracking of moving objects in the video sequence that includes three modules: change detection, object edge extraction and object tracking.(1) At the period of change detection, the higher order statistics theory is analysed, which is combined with the gray difference between two frames. Against the noise video sequences of unchanged background. A change detection method has been proposed, and this method could extract VOP exactly from the key frame.(2) At the period of edge extraction, wavelet transform and muti-resolution theories are studied. Then a wavelet transform multiscale edge detection method is proposed. The experience result proves that this method is a real-time method, and it is better than the other methods in the area of noise-suppression.(3) The use of Housdorff distance in image matching is analysed, and an improved tracking strategy is proposed. In this strategy, an improved Housdorff tracker is used to track the target in the near area of the target. Thus, the time of tracking object would been shorter than that of tracking object at the all area, and the moving object would be extract according to the tracked module. The experience result shows that this strategy speeds up the matching process and improves the resistance to noise. With the research of segmentation and tracking of video object, we take note of algorithms' applicability and relation between segmentation and tracking.in the strategies, author add preprocessing and postprocessing to ensure the validity of the algorithms.