Identification and verification of person by gait features for multi and variant views using artificial neural network

Abstract

In the second phase, front view gait recognition using principal component analysis 15 performed and achieved recognition rate of 97.50 as compared to side view with recognition rate of 67.50.For multi-view gait recognition fusion based approach is presented, where various views of gait images are fused and principal component newlineanalysis is applied for feature reduction, the results shows the improvement in newlinerecognition rate. The CASIA B gait database is used for all the work. newlineAt last in third phase, we have shown the importance of body parts in gait recognition, newlinefor that we have divided the body in three parts upper, middle and lower part and newlineextracted two features for 14 subjects. From the analysis of all the parts, it is revealed newlinethat the lower part has more significance (RR 99.806) than middle part (RR 98.987) and upper part (RR 88.75) has least significance, but when complete body was considered the recognition rate outperforms with the recognition rate of 99.865 for training and 99.2 for testing respectively.