基于彩色图像的非结构化道路检测

提出一种基于彩色图像的非结构化道路检测算法。首先将整幅图像划分为均匀的方块，再依据假设的道路区域提取出的道路特征信息将方块分类；对于包含道路边界的方块区域内部的像素点，采用自动阈值分割方法将其分为道路区域的点和非道路区域的点；最后去除道路内部的斑点以及道路外部的非连通区域。该方法可以避免对所有像素点逐个分析，运算量小，检测速度快，基本满足实时性要求，具有一定的抗阴影和水迹的能力。

Lane and road detection based on colorful image

A new method of structureless road detection based on colorful image was presented. The image was first divided into several small foursquare areas. And then, the characteristic information of road was gotten from the area which is assumed to be the road area. According to these information, the small foursquare areas could be classified. Using the method of self-motion threshold segmentation, each of the foursquare areas which include the shoulder of a road, were segmented into the road area and the other area. Lastly, the spots inside the road and the disconnected areas outside the road were eliminated. This method avoids analyzing each pixel of image. It performs with less calculation and higher speed, and satisfies robust real-time basically. It also resists the effect of shadows and water marks.