Toward texture analysis and feature fusion for GEI-based gait recognition
Gait Energy Image (GEI) is a spatial-temporal representation and it uses the average image to represent the gait pattern. Based on GEI, this paper utilizes the method of texture analysis to describe the local space distribution of pixel brightness in it. Three different local texture features are obtained which are the range of local brightness, the standard deviation of local brightness, and the entropy of local brightness. In order to get a better performance, those features are fused at matching score level with five different rules. The experiments on CASIA database demonstrate that the proposed new texture features are more effective than the original GEI, and also demonstrate that the fusion algorithm outperforms other algorithms.