Recognition Of Human Behavior Based On Conditional Random Field

In video sequences moving human body behavior identification is a task related to video image processing, pattern recognition, computer vision, artificial intelligence and other areas. The final goal of video surveillance is to transplant human vision perception to machine, which enables the video surveillance system to detect and track object, recognize people’s behavior in video sequences. In recent years, though computer vision technologies have improved immensely, researching on a truly robust algorithm of human activity recognition is still a challenging task, because of the complexity of human behavior recognitionIn this thesis, moving human body detection and behavior identification is researched. For the moving objects detection, Gaussian mixture model is initiated by improved average background method, which makes the initial Gaussian mixture model more conform to the background model. When the shadow is detected, the foreground and background color difference method is used so that the target region is extracted contains no shadow.In order to solve the scaling and translating problem of Radon transformation, an improved Radon transformation is used to extract Radon features of motion human for every frame of video sequence. The improved Radon transform is the invariant to translation and scale change. Therefore, normalized processing of size is not needed before feature extraction and motion description, which is more robust and benefits to the following human behavior analysis.In order to solve real-time problem of traditional CRFs(CRF) method, a new CRF method based on principal component analysis method is proposed in this thesis.The metod improves the behavior of real-time identification and accuracy. Experimental results show that this method can effectively carry out human behavior recognition, which not only ensure the recognition rate, but also on the recognition time has been significantly reduced.Traditional recognition of human behavior is researched on the computer, but due to the rapid popularization of mobile applications and Android system is now enormous commercial value. So this article for behavior identification research was conducted on the android system.