摘要
复杂环境下路面交通标识识别是车辆安全辅助驾驶、车辆自主导航等领域的重要研究内容.针对路面交通标识特征,采用一种控制点快速获取方法得到场景透视投影中的变换矩阵,进而在场景重建基础上利用Hough直线检测方法建立当前车道线的区域模型,在该区域内采用最大化一维信息熵方法分割出交通标识,利用Canny边缘检测算子检测其边缘特征,最后通过改进的Hu不变矩特征实现了路面交通标识的有效识别.实验结果表明,所提的方法对于复杂环境下的路面交通标识识别具有良好的可靠性和鲁棒性.
The road traffic signs recognition in open environment is very important in the field of the auxiliary safe driving,the autonomous vehicle navigation and so on.Based on the characteristics of road traffic signs,a control-point information obtaining method is adopted to acquire the trans-matrix of perspective projection.Then,the Hough transform is adopted to detect the regional model of lane line based on scene reconstruction.The road signs are divided with maximum one-dimensional information entropy division and edge features of signs are detected by Canny edge detector.Finally,an improved Hu moment invariants method is adopted to recognize the road traffic signs.Experimental results show that the proposed method for the road traffic signs recognition in open environment has good reliability and robustness.
出处
《大连理工大学学报》
EI
CAS
CSCD
北大核心
2012年第6期908-913,共6页
Journal of Dalian University of Technology
基金
国家自然科学基金青年基金资助项目(51205038)
教育部人文社会科学研究青年基金资助项目(12YJCZH280)
关键词
图像重建
一维熵分割
HOUGH变换
CANNY算子
HU不变矩
scene reconstruction
one-dimensional entropy division
Hough transform
Canny operator
Hu moment invariants
