Fall is kind of major cause of death among elderly people. As older people population rises, detecting fall is important in saving lives. Detecting unnatural fall down among older people living alone at home is an important need. Fall injury is common, and an immediate medication is necessary often. There are three different methods in fall detection, They are ambient sensor, wearable sensor and vision-based methods. This paper explains a real-time video-based fall detection to support the older people through analyzing rate of change of motion with respect to ground point. Our aim is to provide an efficient method to detect fall, without having any physical devices personally. The proposed method is a combination of ground point estimation based on texture segmentation using Gabor filter and calculating the rate of change of angle. A persons movement is tracked by using a Kalman filter method and calculates the angle between the tracked points with respect to a ground point. For experimental analysis, we used two public datasets and analyzed parameters.
Volume 12 | 07-Special Issue
Pages: 232-239
DOI: 10.5373/JARDCS/V12SP7/20202102