Smart camera networks: An analytical framework for auto calibration without ambiguity

Abstract

With the proliferation of smart environment, smart multi-camera networks assume growing significance. Specifically, non-intrusive calibration of such camera networks becomes imperative in smart applications such as telepresence systems, where multi-view imaging/recording needs to be performed in a dynamic setting with continuously changing intrinsic and extrinsic camera parameters. Unfortunately, popular auto calibration methods are known to introduce ambiguity or require manual intervention. In this backdrop, we propose a three-camera configuration (which can be generalized) with a stereo pair having known baseline distance and an additional (mono) camera positioned arbitrarily, and analytically establish the uniqueness of auto calibration in the proposed configuration.