A correlation based face verification scheme designed for mobile device access control: From algorithm to Java ME implementation

Abstract

In this paper, we investigate face recognition based access control approaches dedicatedly designed for camera-equipped mobile devices. We developed a Hierarchical Correlation based Face Verification (HCFV) scheme which is deployable for the most resource constrained mobile devices such as low-end mobile phones with available memory size less than 1MB. The novelty of HCFV is to perform the Partial Correlation Output Peak Analysis (analyze the relations among multiple cross-correlation peaks which are generated from selected sub-regions of a face). The experimental results on the public domain database Yale Face Database demonstrate that the proposed scheme has achieved better performance than using conventional correlation based schemes directly on either full-size or downsized images. Furthermore, in order to investigate the applicability and implementability issues, we have also implemented HCFV on the Nokia S60 CLDC emulator using Java ME programming technology. The experiment shows that HFVC consumes approximately 1/10 memory, 1/6 storage space, and 3.5% processing time in comparison with the result using the conventional correlation methods.