摘要：In order to expand the application of virtual reality technology in network, game, agriculture, industry and other fields, the generation and reconstruction of virtual animation has been deeply studied. However, under the impact of virtual human randomness and human behavior collision, there are the challenges of reliability and real time in the virtual animation. First of all, the model of human behavior is completed by the cooperation of scene perception, brain control and body movement. The initiation, process, and the depth of human behavior are fused to a behavior perception system. Then, the independent sensing vector, the random behavior response capture vector, and the random collision behavior detection vector are deployed in each position of the random person body. At the same time, the random human behavior collision prediction and coordination indicator vector are deployed, and the vector deployment architecture and animation generation system are proposed. Finally, the simulation experiments of MATLAB and visual C++, proved the superiority of the proposed virtual animation collision reconstruction algorithm in prediction accuracy, execution time, virtual behavior judgment accuracy and excellent performance of virtual animation reconstruction effect etc.
