Constrained optimization for opportunistic distributed visual sensing

Distributed networks of dynamically controllable pan-tilt-zoom (PTZ) cameras have high potential utility for tracking and high-res imaging of targets-of-interest maneuvering within a surveillance area. The actual utility that is achieved is determined by the real-time selection of the networked camera PTZ parameters to collaboratively achieve these objectives. This paper proposes a control mechanism for such a network to obtain opportunistic high-res facial imagery via distributed constrained optimization of PTZ parameters for each camera in the network. The objective function quantifies the per camera per target image quality. The tracking constraint that defines the feasible PTZ parameter space is a lower bound on the information about the estimated position for each target. All cameras optimize their PTZ parameters simultaneously using information broadcast by neighboring cameras. At certain time steps, due to the configuration of the targets relative to the cameras, and the fact that each camera may track many targets, the camera network may be able to achieve the tracking specification with remaining degrees-of-freedom that can be used to obtain high-res facial images from desirable aspect angles. The challenge is to define algorithms to automatically find these time instants, the appropriate imaging camera, and the appropriate parameter settings for all cameras to capitalize on these opportunities. The solution proposed herein involves a Bayesian formulation (for an automatic trade off of objective maximization versus the risk of losing track of any target), design of aligned local and global objective functions and the inequality constraint set, and development of a Distributed Lagrangian Consensus algorithm that allows cameras to exchange information and asymptotically converge on a pair of primal-dual optimal solutions. This article presents the theoretical solution along with simulation results.