Frontal-view gait recognition by intra- and inter-frame rectangle size distribution

摘要

Current trends seem to accredit gait as a sensible biometric feature for human identification, at least in amultimodal system. In addition to being a robust feature, gait is hard to fake and requires no cooperationfrom the user. As in many video systems, the recognition confidence relies on the angle of view of thecamera and on the illumination conditions, inducing a sensitivity to operational conditions that onemay wish to lower.In this paper we present an efficient approach capable of recognizing people in frontal-view videosequences. The approach uses an intra-frame description of silhouettes which consists of a set of rectanglesthat will fit into any closed silhouette. A dynamic, inter-frame, dimension is then added by aggregatingthe size distributions of these rectangles over multiple successive frames. For each new frame, theinter-frame gait signature is updated and used to estimate the identity of the person detected in thescene. Finally, in order to smooth the decision on the identity, a majority vote is applied to previousresults. In the final part of this article, we provide experimental results and discuss the accuracy of theclassification for our own database of 21 known persons, and for a public database of 25 persons.