Research And Design Of Wearable Body Posture Monitoring System

Body posture monitoring widely used in medical rehabilitation, health surveillance, video effects, virtual reality and natural human-computer interaction. Existing body posture monitoring can be divided into two ways, non-wearable and wearable. Non-wearable body posture monitoring system can be used when it is not contact with user, so it will have no influence to users’ daily life, but its monitoring range is constrained. Wearable body posture monitoring system can solve this problem well. Existing wearable body posture monitoring system needs so many nodes, and its endurance power is very poor. Therefore, designing a new body posture monitoring system which is versatile and handy is very significant.This thesis has designed a wearable body posture monitoring system which is based on MEMS sensors, and proposed an recognition algorithm for this wearable posture monitoring system.First, for the problem that wearable monitoring systems of the body posture monitoring need more nodes, through simplifying the existing monitoring model of body posture, the human model is established which has four degrees of freedom of upper and lower extremities, and according to the model, the schemes of grouping with nodes and layout of monitoring nodes which are used to obtain the parameters describing the model, are determined, then the hardware scheme of wearable body posture monitoring system based on the combination of integrated accelerometer, magnetometer and barometric altimeter is designed, and sensors were tested in order to ensure their effectiveness.Then, through the mathematical analysis shows that the four degrees of freedom model and five nodes of posture recognition is workable, Presented body posture recognition algorithm based on the wearable body posture monitoring system.It is proved by using mathematical formulas that through the selected scheme of grouping with nodes, data required for gesture recognition can be obtained accurately, and the method of gesture recognition is deduced, thus the feasibility is verified to achieve to monitor posture for the human model concluding four-DOF of the upper and lower extremities through five nodes.Finally, using the body posture monitoring system and its algorithms to test a human-specific posture recognition. Experimental results showed that the system can monitor the body posture effectively.