摘要
本文立足机器视觉,对智能电动车单目视觉的车道线图像检测和识别算法进行研究,主要包括原始图像获取、IPM逆透视图像的转换、滤波降噪(A)、霍夫直线检测、RANSAC和贝塞尔样条曲线拟合和反逆透视变换。实验测试结果表明,本文所改进的车道线检测和识别算法比较准确稳定,对于车辆的遮挡、树木的阴影、不清晰的车道线、明显弯曲的车道线等典型道路情景能够准确检测出车道线,为智能电动车在园区道路环境下自主行驶提供了可行性。
Based on machine vision, the improved lane detection algorithm has been studied with the processing of the image, IPM view conversion, filtering, Hough linear detection, RANSAC and Bessel spline curve fitting and inverse perspective transformation. The test results show that this improved lane detection algorithm is stable and accurate, and can accurately detect the lane line, for vehicle occlusion, shadows of trees, uncleared lane, obviously curved lane and other typical road conditions, providing the feasibility of intelligent electric independent car driving in the park environment.
出处
《计算机科学与应用》
2020年第6期1139-1149,共11页
Computer Science and Application
关键词
智能电动车
车道线识别
算法
机器视觉
Intelligent Vehicle
Lane Detection
Algorithm
Machine Vision
