Identification and analysis of human pose in video

Abstract

Human figure identification is always a challenging move in field of pattern recognition. This paper presents a complete algorithm to find a single object (human body) and identify the object as human being. The algorithm starts the segmentation process with basic frame difference method and use morphological operators, edge detection, feature point generation and finally spline interpolation to find the human like object. After completion of successful segmentation the algorithm takes a gentle approach to identify the object as human body and detect the pose by matching the templates. This paper describes every single step of single human body detection with perfection and ready for real life use.