Activity recognition I

Abstract

Human being with mature brain have powerful ability in observation, recognition and understanding of the biological movements especially human actions. With the assistance of eyes to capture the visual information, the human can recognise a group of people are dancing or a boy is stretching himself. Howbeit, it seems difficult for a computer to understand human activities. The purpose of this project is to develop a system which can parse a image sequence containing a human activity and identity the activity. The system employed the Motion History Image method, an template matching approach, as the two dimensional description of the human activity. The system also makes use of Local Binary Pattern operator to establish a responsible measurement for training and recognition processes. The project mainly incorporate the two method to implement a robust system for single person activity recognition.