1. Introduction
Gait is the overall pattern of bipedal locomotion of an individual. It is largely determined by the person's body structure and behavioral patterns acquired over an extended period of time. Over the past 15 years or so there has been lot of interest in exploring the biometric potential of this pattern as captured through various sensors. Most of the exploration has been with the capture of gait through cameras at a distance. A review of the reported performances [17] show that identification rates range from 80% to 90% for gait video. Recognition rates across the camera view, shoe and speed changes fall, but not by much. Recognition across walking surface-type change, carry condition, clothing and elapsed time appear to be difficult problems. Gait recognition rates from video seem to hold as the dataset size is increased from hundreds to thousands.