A Review of Vision-Based Vehicle Detection and Tracking Techniques for Intelligent Vehicle

Abstract

Vision-based vehicle detection and tracking techniques is of great importance to reduce vehicle collision accidents and increase the driving safety on road. This paper presents a comprehensive review of latest techniques for vehicle detection and tracking. In hypothesis generation of vehicles, motion-based, knowledge-based and stereo-vision based methods are introduced. Hypothesis verification includes template-based, appearance-based and multi-features fusion methods. In addition, three main algorithms are introduced in vehicle tracking. Finally, existing problems and future research directions of this field are summarized.