Research And Implementation Of Human Motion Similarity Calculation Based On Inertial Sensor

With the improvement of science and technology and people's living standard, the research on human movement has been deepening day by day. The study of human action is one of the important aspects, not only significant for human health, in the animation industry, the military field also played a catalytic role. Especially the microprocessor and artificial intelligence algorithm development, more accelerated the human body movement related technology wand. In this paper, based on the existing technology development, the similarity calculation of human motion has been studied in depth. The data are transmitted to the server by serial port communication, and the data are processed by the data processing-related algorithm on the server side. The data are extracted, processed, extracted, and characterized by the data processing.Selection and other operations, and finally through the machine learning classification model to complete the similarity between the different actions of the calculation.In this paper, the background, development status and significance of human motion research are introduced. Then the research on human skeletal model, classification of human action, and representation of human body movement are made from the aspects of human body structure.Based on the research of the human body structure, the human motion capture system is designed, and the system acquisition module, control module and osal operating system equipped with the system are introduced. And use the system to collect human motion data, in the unity3d system to simulate the human action to verify the accuracy of the collected data.The characteristics of the motion data are analyzed and analyzed on the server side, and the data are preprocessed by means of mean smoothing, least squares polynomial fitting and data normalization respectively. The statistic feature is extracted from the action data by statistical knowledge, and the similarity calculation method of data subsequence is proposed to extract the data feature.Finally, the lifting tree classification model and the training data set used in this paper are introduced. Experiments on human similarity calculation are designed and implemented. The experimental results are analyzed from the classification model and human action characteristics.