Design And Implementation Of Wearable Human Behavior Recognition System Based On IMU

In recent years, with the rapid development of the Inertial Measurement Unit and Wireless Body Area Network, and the maturity of pattern recognition theory,Based on the wearable technology of human motion behavior recognition gradually get the attention of researchers, becoming the research hotspot in this field.Compared with the recognition method of image recognition, behavior recognition based on motion sensors shows the advantages of low power consumption, good portability and low cost, in the medical rehabilitation, human-computer interaction,virtual reality and other fields have been widely used. This paper does research in the human recognition method based on wearable multi-sensors,which correctly achieve the recognition of the daily human behavior movement patterns. The main research includes the following aspects:(1) On the basis of existing Inertial Measurement Unit,this paper design a wearable human behavior recognition system, which consists of microprocessor,three-axis accelerometer, three- axis gyroscope, power module and so on, the system can provide real-time, continuous human motion information (acceleration and angular velocity information) to the Android upper computer, on the platform of Android, it can realize the real-time receiving, dynamic display and storage of human motion information.(2) The advantages and disadvantages of the attitude angle calculation based on the accelerometer and the gyroscope are analyzed and compared. owing to the disadvantages of low precision and poor stability of the traditional human body attitude calculation algorithm, this paper presents a method that the acceleration method calibrates the quaternion method for solving attitude angle, the acceleration and angular velocity data are fused to calculate the human body attitude angle in real time accurately.(3) This paper uses the time-domain and frequency-domain analysis method to analyze the human motion information, aiming to recognize daily human behaviors.Based on human motion data acquisition experiment, the characteristics of time domain, frequency domain and attitude angle of human motion data are taken as the characteristic parameters to build the multi classification behavior recognition algorithm based on support vector machine, and using the proposed algorithm to identify human motion patterns. The experiments show that the system can effectively realize the accurate recognition of human daily behaviors.