一种阴影及破损车道线识别方法研究

为了满足车道线识别算法在车道线存在阴影遮挡、破损及污迹覆盖情况下的适应能力,提出了一种新的、有效的识别算法。将原始道路图像灰度化后,采用中值滤波去除图像采集过程中引入的噪声。利用对称局部阈值分割算法对去噪后车道线进行特征提取;并将提取结果与经典分割算法进行对比分析。基于提取出的车道线特征点的分布规律,提出应用改进的RANSAC算法进行车道线识别。分别对在普通公路和高速公路上所采集的视频图像进行实验测试,结果表明,当车道线严重破损、完全被阴影遮挡以及被大面积污迹覆盖的情况,识别算法都能准确地将其识别。

Research on recognition method for shady and broken lane

In order to meet the requirements of the adaptability of shady, broken and stained lane mark identification, this paper proposed a novel and effective lane mark identification algorithm. Turning the color image into gray scale and filtering out noise by median filter were introduced firstly. Then, it extracted feature of the lane by using the method of symmetrical local threshold segmentation and the result was contrasted with the classic segmentation method. Lastly, considering the distribution of feature points, it put forward the identification algorithm based on the improved RANSAC algorithm, and the validity of which was verified experiments using several videos, which were collected from common road and highway. The results indicate that, even for the lane mark which is blocked by shadows completely, broken seriously or covered by a large area of stains, it can be recognized accurately by using the improved RANSAC algorithm.