Fall detection using motion estimation and accumulated image map

This paper presents a novel fall event detection method based on the motion vector and accumulated image map. The proposed method consists of the following steps; i) human object region extraction based on the background subtraction method, ii) generation energy map by accumulating the detected human object region during a consecutive time, iii) motion vector of human object obtained by optical flow method, iv) detection fall event using the knearest neighbor (kNN) classification with the estimated motion vector and energy map.