摘要
文中提出了一种适合于DSP的基于改进的Hough变换的车道标志线检测算法.算法首先对经典Hough变换的前级处理过程——边缘提取中的边缘图像进行梯度运算,然后利用获得的梯度运算结果来确定后续Hough变换过程中本该360°搜索的边缘方向这个参数,同时在基本不影响车道标志线检测结果的前提下对参数空间进行适当分块,以减少车道标志线检测算法的运算量.实验结果表明:文中的方法不仅能够准确地检测出待检测路面的车道标志线,而且能够实时的运行在DM648DSP硬件平台上,获得了很好的车道标志线检测效果.
This paper proposes a suitable for DSP＇s lane marking detection algorithm based on improved Hough transform,the algorithm first computes gradient of the edge image from the edge extraction which is just the preclass process of the classical Hough transform,and then use the computed gradient results to determine the parameter of edge orientation which should be searched from 360 degree in the next process of Hough transform, further more the algorithm to reduce its computation divided the parameter space into many little blocks in the precondition of no affect lane marking detection results.Experimental results showed that：the method not only can accurately detect the lane marking,but also can run in DM648DSP hardware platform in real-time,and get a wonderful lane marking detection results.
出处
《微电子学与计算机》
CSCD
北大核心
2013年第5期157-161,共5页
Microelectronics & Computer
基金
广东省工业科技攻关计划项目(2005A10101001)
