The Study Of Intelligent Monitoring System Based On The Analysis Of Human Body Postures

Focusing on the application of Intelligent Monitoring System, this paper designs a relevant human posture analysis system. It not only achieves to detect and locate people in a picture, but also achieves to recognize human postures based on the information of human region extracted from above steps. The work in the paper makes three main contributions in the field:Firstly, on the research of human object detection and location, we use color matching algorithm to detect human in a picture. We found a human model and use chrominance information of HSI color space to achieve segmentation of human body region. Then we use morphological image processing and spatial filter processing to improve the quality of binary segmentation image. On this basis, we use the method of comparing the area to locate the human body region more accurately.Secondly, on the research of recognition of human postures, we analyze the characters of four human postures in the experiment and choose four geometrical features of human body region as the recognition basis, then use template matching method and BP neural network classification method to complete the recognition. In the template matching method, we complete the recognition of human postures by comparing test picture with standard sample pictures; In the BP neural network classification method, we use the standard sample pictures to train BP neural network connection weights, then use the trained BP neural network to realize the recognition of human postures.Finally, we design a recognition software of human postures by using image processing and neural network toolbox in MATLAB. Then we test the software by using the sample database which is founded by ourselves and get a good effect.