摘要
为了提高车道识别的实时性与鲁棒性,提出了一种车道识别优化方法。通过建立强鲁棒性的车道模型,对传统的道路图像边缘算法优化处理:图像预处理中充分考虑实用性,采用了加权平均法实现灰度化、中值滤波法降噪处理以及自适应Otsu法二值化分割图像、Canny算法实现边缘检测;基于模型利用极角极径法提取图像的感兴趣区域,改进Hough算法实现有效车道线的识别。最后实验证明,通过与传统方法比较,该算法极大提高了车道识别的效率并且增强了可靠性。
A new optimization method of road recognition was proposed which could improve the characteristic of timeliness and robustness. Through the establishment of strong robustness of the model, optimizing the traditional algorithms of image edge detection: taking full consideration of the practicality in the image pre-processing, image gray processing is realized by using the weighted average method,denoising by the median filtering method and binary by adaptive Otsu method and edge detection by Canny algorithm;Polar Angle pole diameter method is used to extract the interest area from the image based on the model , then improved Hough algorithm in terms of efficiency for lane detection. At last, with the traditional methods to compare ,the method used in the experiment greatly improve the efficiency of lane recognition and enhanced reliability.
作者
陈敬义
方博文
张晓东
CHEN Jing-yi;FANG Bo-wen;ZHANG Xiao-dong(College of Mechanical Engineering, Taiyuan University of Technology, Shanxi Taiyuan 030051, China)
出处
《机械设计与制造》
北大核心
2019年第2期234-237,共4页
Machinery Design & Manufacture
基金
山西省经济和信息化委员会技术创新项目(CX2014-45)
关键词
结构化道路检测
自适应
极角极径
霍夫变换
Structured Road Detection
Self-Adaption
Polar Angle and Radius
Hough Transform
