Customizing MPEG video compression algorithms to specific application domains: The case of highway monitoring

Abstract

This paper describes the design of a MPEG coding algorithm targeted to compress video sequences generated by highway surveillance cameras. The physical constraints of vehicles as well as the geometry of the highway environment projected into the camera determine a relevant base of a-priori knowledge, which allows obtaining considerable process­ing speed up as well as higher compression ratio. This paper models the highway environment in terms of rules and shows how a a customized MPEG coding algorithm can take advantage of such rules to improve processing speed and compression ratio.