Fuzzy logic based human activity recognition in video surveillance applications

Abstract

Automatic fall detection using computer vision is a particular case for real time video analysis, efficient in kindergartens. This paper is focused on the design and implementation of a Human activity analysis system. The multiple cameras sends captured frames to the monitoring system via the local network. Through the use of human silhouette, acquired from a smart camera, a shape representation of the human beings was built in real-time and a fuzzy logic inference system was developed for fall detection. The system also allows tracking and localizing children within an authorized area. The alarm is triggered in case of transgression. Experimental results prove that the fuzzy inference system is efficient.