Framework for illumination invariant vehicular traffic density estimation

Abstract

CCTV cameras are becoming a common fixture at the roadside. Their use varies from traffic monitoring to security surveillance. In this paper a novel technique, using Invariant Features of Local Textures (IFLT) & Support Vector Machine (SVM), for estimating vehicular traffic density on a road segment is presented. The proposed approach is computationally efficient and robust to varying illumination. Experimental results have shown that the proposed framework can achieve high performance than extant state-of-the-art techniques in varying illumination conditions. © 2009 Springer Berlin Heidelberg.