Abstract
Due to the increasing interest in human identification by computer vision researchers, the concept of gait recognition has been proposed. This method would allow them to identify people based on their walking gait. In this project, a simple and efficient algorithm has been proposed. The proposed algorithm involves using a combination of methods to track and segment the walking figures. First, a simple correspondence method is used to segment the images from each sequence. The eigenspace transform is then applied to the distance signals from the images to reduce their dimensionality. The proposed algorithm for gait recognition is based on supervised classification techniques. It takes into account the transitional and structural characteristics of walking. In the tests conducted, the proposed algorithm performed well and exhibited low computational costs.