CAJ | 학술논문

为解决结构化车道线识别算法中存在的约束条件多，计算过于复杂等问题，提出一种基于形态学的车道线快速识别算法。该算法首先对道路图像进行中值滤波， Sobel算子增强边缘，利用直方图特征分割图像，然后划分感兴趣区域，接着构造具有车道特征的形态学结构元素去提取车道线，最后概率霍夫变换拟合车道线。实验对比结果表明，针对结构化道路，该算法简单有效，计算量小，具有良好的实时性。
위해결결구화차도선식별산법중존재적약속조건다，계산과우복잡등문제，제출일충기우형태학적차도선쾌속식별산법。해산법수선대도로도상진행중치려파， Sobel산자증강변연，이용직방도특정분할도상，연후화분감흥취구역，접착구조구유차도특정적형태학결구원소거제취차도선，최후개솔곽부변환의합차도선。실험대비결과표명，침대결구화도로，해산법간단유효，계산량소，구유량호적실시성。
In order to solve the problem in the structured lane recognition algorithm which includes many constraints and complicated calculation, this paper presents a fast lane detection algorithm based on morphology. Firstly, this work gets pre-process in road image by handled Median filter. Sobel operator is to enhance edge. A characteristic histogram threshold method is adopted to segment images. And region of interest is disposed with the detection results. Secondly, by constructing morphological structure elements with the feature of lane start to extract the lane line. Finally, Hough transform have been fitting with the best line of lane. The contrast experiment shows that algorithm provides a high real-time, less calculation and simple and effective approach to structured lane.