Research On Real-time Human Detection And Motion Analysis Of Intelligent Surveillance

Video human motion analysis is an important research area which combines various techniques. The focus of the research is in detecting and tracking figures in monocular or multi-ocular videos, capturing and reconstructing the human motion, and then describing and interpreting the human behavior. This research has various applications in human animation, computer game, virtual reality and augmented reality, human-computer interaction video surveillance, sports video analysis, computer-aided clinical diagnosis, and so on.The research of this paper focuses on the detection, segmentation and motion analysis of the moving human in the image sequence captured from a fixed camera. There are mainly two steps in our research. In the first step, an effective connected components labeling method called Connected Chips Linking (CCPL) is proposed in this paper, which is very effective in blob extracting, interference eliminating and contour detecting. After blob extraction, we implement a Mean-Shift search to segment the human body, which could solve the "fragment" problem caused by the interference of the background. In the second step, normalized Radon transform is proposed to extract and organize the shape feature of the moving human body. With the labeled (positive or negative) Radon matrixes as feature pools, an on-line AdaBoost feature selection is implemented to obtain the strong classifier combined by the selected weak classifiers. Finally with the methods described above, we design a real-time aggressive motion detecting system. Experimental results show that our system performs well.