摘要
基于视觉的道路检测是高级驾驶员辅助系统(Advanced Driver Assistance Systems,ADAS)的核心技术。针对空间线模型(SPatial RAY feature)对道路宽度适应能力弱,时间复杂度高的不足,提出了一种改进的空间线模型,利用基础分类器得到的置信度图提取二值SPRAY特征,引入帧间信息复用机制来提高道路区域检测的效率。大量结构化和半结构化道路图像的检测实验证明了该方法能够有效提高道路检测的精度,同时能提升空间线模型对不同宽度道路检测的鲁棒性。
Vision based road detection is a key technology for Advanced Driver Assistance System(ADAS). This paper proposes a novel road detection approach, the core of which lies in the design of the binary spatial ray feature which combines both road appearance and spatial information, and an efficient road classifier based on the extracted feature. The approach mainly consists of multiple base classifiers, binary spatial ray feature extraction and road classification. Moreover, the inter-frame coherence between adjacent images is also applied to accelerate detection speed. Experiments on a large number of structured and semi-structured road images show that the proposed approach can effectively detect road under challenging conditions and is robust against changes in the width of roads.
出处
《计算机工程与应用》
CSCD
北大核心
2018年第6期161-167,共7页
Computer Engineering and Applications
关键词
道路检测
二值空间线特征
高级驾驶员辅助系统
road detection
binary spatial ray feature
advanced driver assistance system