Intelligent Monitoring System Based on Hi3531 for Recognition of Human Falling Action
Dongmei Wu, Hengheng Hao, Li Wang · 2018
The construction of a smart city is the trend of urbanization, and it is bound to unstoppable that the traditional video monitoring system is replaced by a of intelligent monitoring system with specific functions. In this paper, a behavior recognition method based on spatiotemporal shape feature is adopted, and the algorithm is transplanted into embedded chip Hi3531 to realize the real-time alarm for human falling action. Firstly, we extract foreground by the addition of Gaussian mixture background modeling method and three inter-frame difference method. And then, in order to the extraction of feature and the recognition of falling action, we calculate the Hu moment feature twice from Motion Energy Image of motion video sequence, combining with the naive bayesian classifier to classify and recognize. Board-level experiments show that this system can realize real-time alarm when a man fall down.