Arm Motion Recognition And Application Based On Sensor Data

In the past studies, the recognition of the arm movements are more concentrated in the field of computer vision, the use of cameras, smart glasses and other devices to capture the image information to obtain action, and then to identify the image information. And with the development of science and technology, the popularity of sensor devices, action recognition is also more and more the sensor device. The reason is that sensor devices compared to cameras and other equipment has a more convenient, cheaper, resulting in a smaller amount of data and other advantages.In this paper, based on sensor data to recognize arm movements and arm movements according to their complexity can be divided into two categories, one is simple arm movement, another kind is complex arm movements, both simple arm movements or complex arm movements are formed by simple arm movements. According to this feature, this paper defines the arm movements and designs and implements the DTW+HMM hierarchical recognition method to recognize the arm movements.According to the standard DTW+HMM hierarchical identification method,this paper puts forward a number of improvements, which can sacrifice some recognition accuracy, and thus greatly improve the recognition efficiency. So users can choose different solutions according to different application environment, and then achieve the best results. This paper gives an example to illustrate the application of the upper limb exercises and the application of mobile phone protection.Finally, through the experimental results, the effect of the improved scheme on the recognition results of the standard DTW+HMM hierarchical identification method is studied, and the improved DTW+HMM hierarchical identification method is proposed. Through the comparison of the standard DTW+HMM hierarchical recognition method, improved DTW+HMM delamination identification method, the single DTW method, SVM method and DTW+DTW hierarchical recognition method, draw conclusions, need high recognition accuracy applications can use standard DTW+HMM hierarchical recognition method, need efficient identification of the application environment can be used DTW+HMM hierarchical recognition method is improved.