Gait recognition is an unobtrusive biometric, which allows identification of people from a distance by the manner in which they walk. In this paper, a new approach is proposed for extracting human gait features based on body joint identification from human silhouette images. In the proposed approach, the human silhouette image is first enhanced to remove the artifacts before it is divided into eight segments according to a priori knowledge of human body proportion. Next, the body joints which act as the pivot points in human gait are automatically identified and the joint trajectories are computed. To assess the performance of the extracted gait features, fuzzy k-nearest neighbor classification technique is used to identify subjects from the SOTON covariate database. The experimental results have shown that the gait features extracted using the proposed approach are effective as the recognition rate has been improved. © 2011 Springer-Verlag.
CITATION STYLE
Yeoh, T. W., Tan, W. H., Ng, H., Tong, H. L., & Ooi, C. P. (2011). Improved gait recognition with automatic body joint identification. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 7066 LNCS, pp. 245–256). https://doi.org/10.1007/978-3-642-25191-7_24
Mendeley helps you to discover research relevant for your work.