Local image quality metric for a distributed smart camera network with overlapping FOVs

Abstract

A set of camera selection templates, using simple rules based on a local (camera) level metric, are implemented for a twelve camera inward-looking distributed smart camera network. The local metric represents the quality of detection for a given camera node of the target-of-interest and is based on a measurable target parameter. To understand the effectiveness of the camera selections, an analytical framework consisting of a global (system) level metric has been designed. The camera selection methods are able to maintain a desirable global metric performance while using a subset of the total cameras available. This is true even when the system undergoes perturbation by the loss of a single camera or by a single occluding target.