The Swimmers Motion Detection Using Improved VIBE Algorithm

Abstract

this paper proposed a novel method for drowning person detection in the swimming pool using video images. For background extraction and to update the exact motion area from the whole video using frame by frame difference vibe algorithm is used. Static and dynamic features are detected to recognize the normal swimmer and drowning person. The present invention discloses video-based swimming pools drowning event detection method. In the detection process Time of map(Tom), the method is used to improve the traditional VIBE result. The sequence of video images of the swimming pool is collected in real-time by using a camera installed above the water surface, which mainly includes three steps of swimmers detection, swimmers tracking and drowning person behavior analysis. In the aspect of swimmer detection, an improved VIBE swimmer detection algorithm is proposed, and the algorithm is used to determine the swimmer's position. The swimmer tracking and particle filter based on the color distribution model which is combined with the nearest neighbor data association algorithm to achieve tracking of multiple swimmers. In the analysis of drowning behavior, three characteristics of drowning behavior are proposed to determine whether the swimmer is drowning. The invention can monitor the swimming pool in real-time through the camera installed above the water surface in a real public swimming place, and automatically detect the drowning person, which has great engineering application value.