Implementation of a real-time automated face recognition system for portable devices

Automatic face recognition (AFR) systems have been the subject of considerable research over the past thirty years. There are numerous algorithms in the literature that provide a high detection rate. Most of these algorithms are however solely implemented on general-purpose computers and are therefore not suitable for embedded systems. With increased commercial interest in portable devices with advanced real-time user-authentication support, there is a need for more cost effective and low power implementations of AFR. In this paper, we present the implementation of an efficient automatic face recognition system. Features such as realtime execution and moderate memory usage of the proposed implementation, alongside the low-power consumption of the ADI Blackfin processor used as our development platform, makes this implementation very suitable for applications needing real-time user-authentication. The discussion includes how the choice of processor platform, design tools and most importantly the algorithm used in the implementation, could have significant impact on the realization of DSP-based systems.