Human Gesture Recognition Based On Body Movements

Human gesture recognition is an important research direction in the field of computer vision, and is used widely. But the human body movements in time and space are complex and arbitrary motion. Both of them bring a big challenge to recognition. Therefore, how to effectively represent the human body movements, how to use an effective method to identify human action have become a key problem which needs to be solved in the human body posture recognition.In this paper, a new algorithm of motion representation and recognition are proposed, which is based on the approaches studied in previous research. The main research work is as follows:(1) According to discussing for the development in domestic and foreign, and we analysis of human motion capture method. At last, we determine the appropriate collection methods, and establish a sports database;(2) Through the study of existing research results, for human motion data volume, low efficiency in the latter part of the identification problem, we propose a novel motion feature extraction method.(3) By the study of multiple motion recognition algorithm analysis, we propose a typical time warping algorithm and identify a given action;(4) If the results of recognition for the first time are not very satisfied, we can learn through relevance feedback and the system gives satisfactory results through feedback.Through a large number of experimental results and data analysis showed that using of the segmentation, clustering, matching algorithm which we proposed in this paper, the test system can accurately identify the human body posture.