Giseok KIM and Jae-Soo CHO^*

Department of Computer Science and Engineering, Korea University of Technology and Education, Cheonan, Chungnam 330-708, Republic of Korea

(Received May 9, 2012; Revised August 28, 2012; Accepted August 29, 2012)

In this paper, we propose a robust real-time vehicle detection and inter-vehicle distance estimation algorithm for vision-based driving assistance system. The proposed vehicle detection method uses the combination of multiple vehicle features, which are the usual Harr-like intensity features of car–rear shadows and additional Haar-like edge features. The combination of two distinctive Haar-like intensity and edge features greatly reduces the false-positive vehicle detection errors in real-time. And, after analyzing two inter-vehicle distance estimation methods: the vehicle position-based and the vehicle width-based, we present a novel improved inter-vehicle distance estimation algorithm that uses the advantage of both methods. Various experimental results show the effectiveness of the proposed method.

Key words: vision-based vehicle detection, Haar-like features, inter-vehicle distance estimation, driving assistance system

^*E-mail address: jaesoo27@koreatech.ac.kr