Volume 10 número 2
Action-Based Road Horizontal Shape Recognition
Zhencheng Hu and Keiichi Uchimura
This paper presents a new action-based shape model concept for road horizontal shape recognition using monocular color image obtained by vehicle-mounted camera. It classifies the road horizontal shape into straight, curve, curve-in and curve-out four models in order to obtain the simple road shape description relating to the vehicle control actions directly. Contrary to other road shape recognition methods, this approach is simpler, faster and directly relating to vehicle control actions, by taking the driving experiences of human being and the highway construction rules into account. Experiments on synthetic images and real outdoor road scenes show the effectiveness of this approach.
Cópia em PDF (260 kbytes)
Volta ao Índice do Volume 10 .
Disponibilizado em 24/05/1998