The Research Of Human Behavior Analysis Description And Recognition

Video description and identification of human behavior has increasingly become the focus of world attention. With the advance of "smart city" planning,, intelligent monitoring, human behavior description and identification technology industry has started to take shape. Description and recognition of human behavior still exist some bottlenecks, high sensitivity to the environment, computational complexity, to some extent hindered its application and promotion. It also has become the direct driving force of this article.For this reason, this paper selects human behavior description and identification of these two key issues to be studied. First, the paper is divided based on the identification of actors and the research object, which can be from the video frame first locate the body area. This paper is divided into two parts-the description of human behavior and human behavior recognition. Description of human behavior is the focus of the paper.Detection and segmentationa of the body is the first premise of its research. The results of region detection and segmentation affect the follow-up work directly. Behavioral description and analysis of this study is based on the detection and segmentation of the target area on the body conducted. In this paper, we use background subtraction to achieve the human region detection and segmentation. We focuse on the analysis of the contour information of the target region, to extract the human body activity information. Innovative work in this paper focuses on how to use the body’s contour information describing its specific behavior.This paper analyzes human behavior based on freeman chain code description of the algorithm, Studied freeman chain code normalization, poor differentiation, using standardized sequences of human behavior contour inflection point, the core technology of the sequence length sequence and the angle between the inflection points are given to match the cycle of human behavior to give the body contour curves movement cycle behavior. The software design and implementation for the "Walking", "Running", "Boxing", "Handwaving", "Handclapping","Jogging" six movements have been completed. We use SVM extracted feature vector space method to identify. Experiments show that our method is fast and effective.The experiment has proved that the description based on human behavior freeman chain code of human behavior and based on SVM can quickly identify the behavior of motion feature extraction, the extracted features can represent their corresponding behavior effectively,and it can be effectively recognized in analysis of human behavior has a certain reference value.