Research On Human Body Posture Recognition Algorithm Based On Wireless Body Area Network

With the development of modern society and the change of people’s life style,Aging and sedentary youth has become a focus issue. The perennial bedridden old and the office of the white-collar workers in computer are puzzled by some chronic diseases. Due to the fast-paced life and lack of medical resources, chronic diseases and some intermittent episodes of disease are often ignored. A small medical detection and recording device which is capable of carrying and not have movement burden was born. Wireless body area network(WBAN), as the application of wireless sensor network(WSN)in the field of medical treatment, has become a hot research area increasingly. Not only has important applications in medical detection equipment, but also has a wide range of applications in the field of entertainment, military, identity authentication and aerospace. But such an application value of the technical field, there are still a series of technical problems have not been overcome. Especially since the carrier of WBAN is a high dynamic human body, the complex and changeable human body postures make the communication of WBAN more complex. So it has important theoretical value and practical application value to explore the human body posture recognition algorithm and the human body posture signal acquisition system in the wireless body area network. The main work is summarized as the following three aspects:(1) Establishment of human body posture recognition system, in order to establish datasets which is used to test the algorithm of human body recognition. Study and extract the characteristics of human body posture signal in WBAN, compare and choose the suitable method of data preprocess, matrix dimension reduction and human posture recognition in WBAN. Through the study on characteristics of human body posture signal, extraction and selection the characteristic sequence of the human body postures signal. Study on the working mode and signal collection mode of Micro-electro-mechanical Systems sensor(MEMS), select the suitable type of sensor to set up human body posture recognition experiment data collection system. According to the different age and figure of the experimenter, record the raw data respectively. Establish the datasets is used to verify the human body posture recognition algorithm.(2) Hierarchical recognition algorithm of the body posture based on wireless body area network is proposes. The algorithm preprocesses the posture signals, filter out the outliers and sample, extract and select posture signals which have the classification recognition(CR).Recognize accurately 5 kinds of postures using only three layers of determination conditions. Finally, the recognition rate is up to 96.5% by calculating through the simulation experiment. Through the improvement of recognition rate by setting parameters, finding the relationship between the threshold setting rules and the recognition rate.(3) A human body posture recognition algorithm based on BP(back propagation) neural network for wireless body area network is proposes. During analyzing and summarizing shortcomings of hierarchical recognition algorithm of the body posture, finding that the algorithm which extracts and chooses posture characteristic signal needs a large number of human work, so designing an efficient algorithm to reduce the human work. BP neural network is regarded as a human body posture classifier, improving posture recognition rate and network robustness. Through the simulation results, discussing and summarizing the relationship between the network parameters and network structure with posture recognition rate.In this paper, the main work is providing a new approach to solve human posture recognition problems based on WBAN, proposing that BP neural network is applicated in human posture recognition problems of wireless body area network and proven to improve posture recognition rate and increase the body posture recognition algorithm robustness. The two algorithms proposed in this paper do the foundation work for further studying of WBAN.