Multi-based scene analysis and feature learning method for detecting the elderly fall

The present invention relates to a scene based on feature analysis and study of the elderly fall detection method, changes in body posture was added, and the center position of the effective area of ​​the body subjected to the correction ratio of three determination conditions. And real-time learning scenario, the difference between a seating area, which can detect events fall under different monitoring scenarios. The proposed algorithm of the present invention is also the number of false positives fall detection algorithm has provided a corresponding solution. The method of low complexity, low computation and the higher detection rate.