Human gait recognition system based on shadow free silhouettes using truncated singular value decomposition transformation model Online publication date: Sat, 29-Nov-2014
by Rohit Katiyar; Vinay Kumar Pathak; K.V. Arya
International Journal of Artificial Intelligence and Soft Computing (IJAISC), Vol. 4, No. 4, 2014
Abstract: In present scenario biometrics system are getting popular day by day and they are classified based on subject's cooperation and non-cooperation nature. Gait biometrics is one of the popularly traits used for person identification. The gait biometrics has an edge over the other biometrics traits as it works well even if the subject is not cooperative. The multiple problems such as shadow detection, removal of moving subjects in visual surveillance, less number of gallery probes in different conditions and existence of multiple views in the data are encountered due to surveillance camera and affect the performance of the gait recognition system. In this paper, we used three algorithms to overcome these problems up to an extent. In first algorithm, photometric properties based method is used to remove shadows at the time of silhouette generation from the recorded video. Then, an algorithm for synthetic gait energy image (GEI) templates generation employed to increase the corresponding gallery probes dataset and finally, singular value decomposition transformation is applied to transform the gait feature from one view to another. The performance of the proposed algorithm is experimentally validated on a benchmark dataset of indoor as well as outdoor video sequences by comparing it with the existing algorithms.
Online publication date: Sat, 29-Nov-2014
If you are not a subscriber and you just want to read the full contents of this article, buy online access here.Complimentary Subscribers, Editors or Members of the Editorial Board of the International Journal of Artificial Intelligence and Soft Computing (IJAISC):
Login with your Inderscience username and password:
Want to subscribe?
A subscription gives you complete access to all articles in the current issue, as well as to all articles in the previous three years (where applicable). See our Orders page to subscribe.
If you still need assistance, please email email@example.com