作者: Dezhi Gao , Wei Li , Jianmin Duan , Banggui Zheng
DOI: 10.1109/ICAL.2009.5262562
关键词:
摘要: In the paper, a practical method for road detection is proposed. Many features of vehicle are introduced to detect preceding vehicles. According characteristics lane marks and vehicles, improved Sobel operator convolution kernel in image preprocessing enhance robustness algorithm. order be feasible change illumination conditions, adaptive double threshold get binary image, edges which needed during front vehicles extracted with Hough Transformation SUSAN algorithm based on different contrast ratio. apply information implement departure warming system forward collision system, both distances from host left right lines estimated according inverse projection. Experiment results indicate that this conditions can realize tracking efficiency simultaneously.