Centre-of-Mass Based Gait Recognition for Person Identification

Abstract

Background: Gait recognition focuses on the identification of persons from their walking activity. This type of system plays an important role in visual surveillance applications. The walking pattern of every person is unique and difficult to replicate by others. Objective: The present article focuses on to develop a person identification system based on gait recognition. Methods: In this article, a novel gait recognition approach is proposed to show how human body Centre-of-mass-based walking characteristics can be used to recognize unauthorized and suspicious persons when they enter in a surveillance area. Walking pattern varies from person to person mainly due to the differences in the footsteps and body movement. Initially, the background is modelled from the input video captured through static cameras deployed for security purpose. Foreground moving object in the individual frames is then segmented using the background subtraction algorithm. Centre-of-mass based discriminative features of various walking patterns are then studied using Support Vector Machine(SVM) classifier to identify each unique walking pattern. Results: The proposed system has been evaluated using a self-generated dataset containing a side view of various walking video clips. The experimental results demonstrate that the proposed system achieves an encouraging person identification rate. Conclusion: This work can be further extended to provide a general approach in developing an automatic person identification system in an unconstrained environment.