Abstract

Introduction Wehave derived a novel approach to the detection and recognition of human gait. In gait detection, we #nd a spatiotemporal pattern that signals the presence of a walking person. In gait recognition,we seek to identify the individual who is walking. It is known that humans can detect and recognize gait with reduce spatiotemporal sequences #suchasmoving light displays# #3, 10#, and wewould like to give similar capabilities to machines. Any reasonable approach to the interpretation of human motion must impose a model of a human and explain how visual observations are to be #tted to the model. Model recovery is di#cult for a number of reasons. As with most object model recovery, this process should be insensitive to lighting, position, and size. In modeling humans, the recovery process should not be sensitive to clothing or any other features speci#c to a particular individual. Furthermore, unlike most objects, the human body is composed of a large number of parts which can

Citations