Hybrid method for Gait recognition using SVM and Baysian Network

The Gait recognition is the 2nd generation of biometric identification technology which aims to identify people at a distance by the way they walk. Due to the fact that there has been increasing research interest in the identification of an individual in access controlled environments such as the airports, banks and car parks, it has been observed that the effective human gait recognition plays a very important role in such video surveillance based applications. This paper proposes an effective Gait recognition method for automatic person recognition using SVM and Bayesian Network. In this method frames of videos are used as an input, these videos are live and are from the CASIA dataset. The background subtraction is done using Gait Pal and Pal Entropy and a Median Filter is also used to remove noise from the background. Feature selection is done using the Hanavan´s model to reduce the computational cost during training and recognition. Support Vector Machine (SVM) and Bayesian Network are used for training and testing purpose. The experimental results show that the proposed approach has a very effective Correct Classification rate (CCR).