Towards distributed face recognition on self organized storage built from wireless sensor nodes

Abstract

Distributed approaches to face recognition have been widely studied for over a decade. One of the remaining problems with these approaches is that of coordinating the heavy storage requirements in a distributed way. We present a modification of existing facial recognition algorithms deployed on a self organizing, reliable and robust platform composed of wireless sensor nodes that should yield a self-organizing algorithm for face recognition, which manages its storage in an entirely distributed way. Our preliminary results, though tentative, suggest that these expectations are plausible. As an effective facial recognition algorithm, our implementation is still a work in progress; but the modest computational requirements of the storage platform, coupled with the parallelizable operations already in current distributed face recognition techniques should eventually yield a distributed algorithm that can be scaled up to run on a very large system of simple processors.