Automatic identification of person using fusion of gait features

Abstract

Person identification is an essential activity in Surveillance system in order to identify the behavior of the person in the environment. Each person can be identified by his/her appearance and clothing types. But appearance and clothing of a person may vary in different views and in many scenarios. In this paper, a new model has been proposed to identify a person by unique triat (i.e., walking style) through static and dynamic features of gait. Initially, body part model has been build with eleven joints. The static and dynamic features of the gait has been extracted from each person and feature vector has been derived. Stride length, hip to knee distance, knee to ankle distance has been considered as Static Parameters. The joint angles from hip and knee, joint angular velocity and acceleration has been considered as Dynamic parameters. The Euclidean distance measurement has been done. The novelty of this paper lies in feature extraction and codebook generation for each and every person in order to identify the person reliably. Person has been identified in different situations and the results obtained through this method seems to be promising.