Fisheye lens camera based surveillance system for wide field of view monitoring

Abstract

This paper presents a fisheye lens camera based surveillance system for wide area monitoring. The proposed system consists of: (i) digital anamorphic transformation, (ii) two layer panorama generation, and (iii) geometric transform invariant object detection and tracking for the fisheye camera based system. Since peripheral regions of acquired images are not used in the fisheye lens based imaging system, the proposed algorithm transforms a circular fisheye lens image into an anamorphic image using directional interpolation to maximize the usage of lost regions. In detecting and tracking objects, the proposed algorithm adopts a region based approach to provide robust feature points and object shapes against to various tracking challenges including geometric transformation, sub sampling, interpolation, and dynamic illumination. Experimental results demonstrate that the proposed method can successfully detect and track moving objects in the very wide field of view without using multiple cameras or an optical anamorphic lens