Dynamic Human Body Pose Detection Wireless System Based On ADXL345

Dynamic detection of action is a new science. It can be applied to count and recognition of the effective action in sports and fitness shape, such as running, rope skipping, and sit-ups. Its application range is very extensive. Therefore the demand of designing and developing detection system to detect poses of action is very strong. This system covers a small area, has multiple functions, is easy to carry, has high precision, and can aggregate network to summary data.This paper discusses and researches feasibility that detects the specific action and poses using NRF24L01 wireless network technology and system stability. The existing SMV algorithm based on acceleration sensor is used to identify the action and poses. The algorithm has some probability of miscarriage and the parts with high flexibility like arm can be simulated. This paper analysis the movement poses on the human body, and proceeds feature extraction, builds a first-order autoregressive model because of the movement on the human body is cycling movement and has rule to follow. With the assist of the first-order autoregressive model, this paper effectively compresses and controls the measurement range of SMV algorithm, and introduces angle detection to assist the SMV algorithm to distinguish simulation action and the effective action and obtains higher detection accuracy. This paper designs PC software and detection system focus on the improved method of Action detection based on ADXL345 design, and gives the hardware design principle and the improved test results of algorithm. Ultimately develops PC analysis software and the principle prototype of system based on ADXL345 which is wireless and which is used to detect to dynamic pose on the human body, realizes the detection to the specific action pose on the human body.