Extraction of Posture Silhouettes using Human Posture Feature Points and Spatial Masks for Activity Recognition in Public Place

Abstract

Presently, an automated system has been required for public place security. Recognizing human postures in public places has emerged as a global solution for understanding human behavior in public places. In this work, a model to extract a human feature attribute of its posture has been presented to identify human behavior. The research work in this paper focuses on identifying the seating and standing postures of a person. The proposed methodology aims towards extraction of the human attributes from public places using spatial masks. Consequently, in this process, unwanted details from the background have been removed using the technique to focus on human postures only. The feature extraction process gives us blob vector and posture vector to evaluate human authentication and posture apprehension.