Matching Based On The Relationship Between The Parallel Components Of Human Identification

Human body recognition is a hotspot in three-dimensional recognition. How to recognize the human body accurately and quickly is a significant problem. Part-level human body recognition is widely used in production process, visual navigation and virtual reality and so on. It's a practical and efficient scheme of human body recognition. In this paper, we carried on researches for the method of parallel part-level human body recognition based on relational matching. And all of our efforts are to recognize the human body accurately and quickly.Aiming at part-level human body recognition and the characteristics of human joint, this paper deeply analyzed the principle of relational matching and combined structural errors with the gesture of human so as to recognize the part-level human body by using minimal structural errors matching. At the same time, a human trunk restriction strategy was proposed, which is used in interpretation tree search in part-level human body recognition. When the scene is very complex, the scale of interpretation tree will be very large and this strategy can reduce the time spent in search and improve the search efficiency. In order to improve the efficiency of human body recognition, we used the idea of parallel search tree in the interpretation tree search, and analyzed the parallelism among branches, nodes and layers, and designed parallel search tree based on part set division. The method of parallel recognition of human body was built, which can reduce the time spent in constructing and accessing interpretation tree. So this method prompted the efficiency of human body recognition.In addition, we exploited a complete experimental platform used for carrying out experiments of part-level human body recognition. We used the principle of minimal structural errors matching and parallel search tree into the design of the platform, and executed a series of simulation experiments, the results of which were analyzed and compared. Those experiments showed better accuracy and efficiency in recognition of part-level human body.