Contemporary Feasibility of Image Mosaic Based Vehicle Position Estimation

Abstract

Perhaps one of the simplest mechanisms for navigation in an area is to use real-time imagery to track vehicle motion over a very large, previously stored, high resolution image mosaic of the scene. In structured environments, image mosaics of textured surfaces present a potential for inexpensive, highly repeatable position estimation that is both higher performance and cheaper than existing commercial alternatives. This paper presents a case for the feasibility of such an approach using contemporary cameras and computing devices.