Studies Of Video Target Segmentation, Macro-shape Description, And Motion State Detection

The developments of video processing technology are stimulated by their social requirements. In military applications, Infrared imaging guidance technology is becoming the trend of new generation accurate guidance systems. The technology accomplishes the intelligent detection and tracking tasks with image processing techniques. In multi-medium and video communication applications, the limited communication bandwidth impairs the popularization of multi-medium products which operating over the PSTN. One effective method is to use video encoding techniques to compress the video data, especially the second- generation video encoding standards. These encoding schemes can provide both the higher compression ratios and more widely application platforms. In traffic-parameter detection systems, magic loop detectors have been used to estimate the traffic parameters in early visual traffic parameters detection times. Yet such devices have shortcomings of not conveniently to be installed, can't accomplish the high-level detection tasks such as queue length calculation, plate's license number recognition, and individual vehicle tracking etc.. Therefore, novel intelligent video processing techniques are inevitably required to take place of the old detection means. In video surveillance systems, much more video processing techniques are used for their fast, accurate detection results. Since all of above applications are indispensable to use the video processing theories, so it necessary and important to study novel video processing techniques commonly used in those applications. Although there are many different applications, yet all of their realizations need to some same fundamental video processing techniques. For example, in the infrared image guidance systems, noises and backgrounds are needed to be suppressed, motion regions are detected, features are extracted, and tracking algorithms for moving targets are considered. The second generation encoding schemes also need to study many video processing techniques, including the segmentation of background/foreground, textures analysis, features extraction, motion prediction and tracking, visual models study, and syntax analysis. Also, in the intelligent traffic parameters detection and surveillance systems, motion-region detection, features extraction, and motion states tracking are considered. All of the above discussions indicate that it is very necessary and important to study techniques commonly used in most applications. In this paper, series of important video processing techniques are studied. Mainly techniques are, background/foreground segmentation, motion region detection, features extraction, geometry parameters estimation, multi-motion states detection, and motion region tracking. Many novel theories and algorithms are proposed in the paper, they are introduced as follows. As for the motion regions extraction techniques, one commonly used method in most papers is called adjacent difference or optical flow field iterations. Another alternative is consists of two steps, the first step is adjacent difference, the second step is to search the motion regions in the frame being processed. The first scheme will result in overlapping phenomenon, and the second one is not available in complex background or in a scene where the processed targets are similar each other and very near. In this paper, mathematical models of motion regions about the adjacent frame difference are proposed and discussed. In order to reduce the errors resulting from the adjacent difference, we adopt the video sequence hier...