A Pervasive Computing Surveillant System Based On Human Behavior Analysis

In recent years, security problem is getting more important in life and industry, and the research on intelligent surveillance system has attracted more and more attention. With the advent of pervasive computing technology, security surveillant system in intelligent environment will be studied and developed. The system can detect activities in environment and provide alert message, then user may receive them through Internet. At the same time, the user may get information by telephone and mobile phone from PSTN and CMCC respectively.We research and design an intelligent pervasive computing security system based on human behavior recognition in the paper. The system can detect and recognize human behavior with vision technology, and send out alert information through pervasive computing network. Generally, user may receive these messages by mobile phone, telephone and computer. Six simple human behaviors can be recognized which are walking, standing, sitting, squatting, leaning and lying. The innovative researches include:1) To design and implement an intelligent security system based on Human-Centered Pervasive Computing system model which is proposed in the view of engineering.2) To recognize human behavior with PCA algorithm. K-L transform can compress dimension of eigenspace, and we use the low-dimension eigenvector instead of original eigenvector to match unknown posture.3) To process every images through extracting posture feature in order to reduce affect of different dressing. We detect moving human contour with Canny algorithm and make use of processed images to build eigenspace.Experimental results show that the system provides an efficient way to detect and recognize human behavior, and the ratio of correct recognition is high. It also sends out alert SMS message to user's mobile phone well. So, the security system will have real value in a long run.