Authors: P. Jia, H. Hu
Addresses: School of Computer Science and Electronic Engineering, University of Essex, Colchester, CO4 3SQ, UK. ' School of Computer Science and Electronic Engineering, University of Essex, Colchester, CO4 3SQ, UK
Abstract: Recently, some novel human-machine interfaces (HMI) have been created for disabled and elderly people to control intelligent wheelchairs (IW) using facial and head gestures. To operate a wheelchair in this new visual-based control mode, user identification should be conducted beforehand. Rather than traditional user identification that requires the user to input his/her username and password by typing, the state-of-the-art biometric-based user identification provides a more convenient way for the disabled users. This paper first elaborates active shape model in detail; then, video-based user identification using Mahalanobis distance is presented. As an extension, an adaptive learning module is designed to append or update the user|s face record in the constructed face database. Experimental results show that our login subsystem is able to function well for a comparatively small face database.
Keywords: intelligent wheelchairs; user identification; human-machine interface; HMI; active shape models; ASM; disabled users; elderly users; wheelchair control; facial gestures; head gestures; wheelchair operation; visual control; biometrics; adaptive learning; face database; login subsystems.
International Journal of Advanced Mechatronic Systems, 2009 Vol.1 No.4, pp.299 - 307
Available online: 06 Jun 2009 *Full-text access for editors Access for subscribers Purchase this article Comment on this article