Vision Based Surveillance System for Detection of Human Fall

The intention of this paper is to present a novel method for real-time detection of human fall from the real-time video which is taken from the static digital camera which is fixed in the indoor that provides a secure environment and to improve the quality of life of the old person, children, patients and elderly. The proposed work is based on two techniques, an Ellipse approximation and Motion History Image (MHI). The novel work includes removal of shadows for best detection of human in an indoor environment. In this work human fall detection by considering ellipse approximation, Motion history image and combining both the techniques. Results were compared all techniques for the different possible position of human and it shows that the combined technique gives better accuracy and efficiency of human fall detection compared individuals techniques.