基于视觉的缩微车车道线检测

Detection of Lane Lines for Vision-based Micro-vehicles

针对缩微车车道线检测算法鲁棒性和实时性差的问题,利用一种数学形态学与最小二乘法相结合进行缩微车车道线检测.首先对采集的道路图像进行预处理,用数学形态学滤波算法去除室内光线变化对车道线图像的影响,运用大津算法和Canny算子分别对去除光照后图像进行分割及分割后图像的边缘检测,最后采用霍夫变换和最小二乘法相结合的方法检测缩微车车道线.在室内强光照等复杂环境下,能准确快速检测出缩微车车道线,解决了传统缩微车车道线检测算法鲁棒性和实时性差的问题.

Due to the problem that the micro car lane mark detection algorithms are poor in robust adaptation and real-time performance,an improved method combined mathematical morphology and least squares method is used to realize the micro car lane mark detection.Firstly,road images are preprocessed,and the influence of indoor lighting changes to lane images is removed by the improved morphological filtering algorithm.And then,the images are dividing by OTSU threshold dividing algorithm,and after dividing,Canny edge detection operators are used to realize edge extraction.Finally,the cumulative probability of the Hough transform combined with least squares method is used to realize the micro car lane mark detection.The experiments show that the proposed algorithm can detect the micro car lane mark in strong light or lack of light and other complex lighting conditions quickly and accurately,which solves the problem that the tradition micro car lane mark detection algorithms are poor in robust adaptation and real-time performance.