This project will propose a study of a smart home system based on the anthropocentric video surveillance and analysis. The project is designed to complete the major development issues in three years. In the first year, we will develop coding/decoding techniques for video compression and realtime streaming transmission technology for remote surveillance. An object detection, segmentation, and classification method will be proposed for human body detection and recognition. In the second year, we will develop human behavior feature extraction and analysis, human body tracking, and abnormal behavior recognition techniques to detect and recognize human behavior and convert to abnormal behavior information. The abnormal behavior information can be used to integrate and control the smart home system and build a smart home abnormal behavior feedback system. In the third year, we will focus on the personal privacy issues and develop data hiding and video encryption techniques to protect personal data security. We will integrate all the proposed methods and techniques to complete the anthropocentric video surveillance and analysis based smart home system. The results of video analysis and abnormal behavior information produced by the proposed system will be used for the study of the association rule between the depressive level and the abnormal behavior and emotional change in the first sub-project, for the study of emotion recognition in the second sub-project, for the cloud database information/knowledge retrieval of the melancholia in the third sub-project, for the integration of body action detection in the fourth sub-project, and for the development of an intelligent wheelchair care system in the sixth sub-project. Finally, this project will help to complete the self-care-oriented mobile cloud melancholia medical knowledge mining system.