Study On Method Of Video Objects Segmentation Based On Spatio-temporal Information

With the rapid development of information technology and Internet, multimedia has become the main media that people get information. The efficient impression and processing methods are needed to deal with the information, such as store, transmission and index. Well, the video processing technique based on objects is one of the important means. Video object segmentation is the base of video processing technology. As a key supporting technique for video processing, video object segmentation research has academic significance and application importance.The existed methods of moving objects segmentation in video sequences are analyzed in this dissertation. An algorithm is adopted for segmentation of video objects based on spatio-temporal information to solve the questions that the object locations in spatial segmentation and object edges in temporal segmentation can't be confirmed. That is the method that combines edges detection with motion information. The temporal segmentation, the spatial segmentation and the post-processing of video objects are focused.Firstly, for temporal segmentation, motion regions are extracted according to the median (MED) and the median absolute difference (MAD) of the difference image. This method is feasible and can quickly determine the motion regions. Then, SUSAN and watershed transform are used for spatial segmentation to find edges of images on this basis. At last, combining temporal and spatial segmentation can remove the redundant part of motion regions. An exact segmentation result can be obtained.Plenty of simulation experiments have been carried out by this method. And results show that this method is simple and feasible. It has greater segmentation speed and better results.