A Modified Walk Recognition System for Human Identification Based on Uncertainty Eigen Gait

Abstract

Gait based recognition is one of the emerging new biometric technology for human identification, surveillance and other security applications. Gait is a potential behavioral feature to identify humans at a distance based on their motion. The use of new methods for handling inaccurate information about gait features is of fundamental important. This paper deals with the design of an intelligent gait recognition system using interval type-2 fuzzy K-nearest neighbor (IT2FKNN) for diminishing the effect of uncertainty formed by variations in energy deviation image (EDI). The proposed system is built on top of the well-known principal component analysis (PCA) method that is utilized to remove correlation between the features and also to reduce its dimensionality. Our system employs IT2FKNN to compute fuzzy within and in-between class scatter matrices of PCA to refine classification results. This employment makes the system able to distinguish between normal, abnormal and suspicious walk of a person so that an alarming action may be taken well in time. Interval type-2 fuzzy set is involved to extend the membership values of each gait signatures by using several initial K in order to handle and manage uncertainty that exist in choosing initial K. The result of the experiments conducted on gait database show that the proposed gait recognition approach can obtain encouraging accurate recognition rate.