Vision-based approach for people tracking using gait in distributed and automated visual surveillance

Abstract

Recent research studies have now confirmed the possibility of recognizing people by the way they walk i.e. their gait. As yet there has been little formal study in surveillance systems for identity tracking using gait signature over different camera views. We present a new approach for people tracking and identification between different non-intersecting un-calibrated cameras based on gait analysis. A vision-based markerless extraction method is being utilized for deriving the gait kinematics as well as anthropometric measurements in order to produce a gait signature. Given the nature of surveillance data, a parametric Fourier descriptor is being used to guide the extraction process of the legs. The novelty of our approach is motivated by the recent research for people recognition using gait. The experimental results confirm the robustness of our method to extract gait features in different scenarios with a classification rate of 92% for lateral views. Furthermore, experimental results revealed the potential of our method to work in real surveillance systems to recognize walking people over different views with achieved cross-camera recognition rates of 95% and 90% for two different views.