路面检测对于自动驾驶系统具有极其重要的作用,其具体的应用方面包括检测辅助、避障、自动导航等。基于视觉的路面检测主要就是对图像中每一个像素点进行分类,区分其是否为路面。到目前为止大部分的路面检测算法是应用于白天。在本文中,我们集中解决夜间的路面检测。我们利用一个近红外摄像头来采集夜间图像。检测时,首先利用平面反射模型来对图像中的路面部分进行拟合,然后,一个基于像素点的分类方法被用来对图像中的每一个像素点进行分类。在实验部分,我们将我们的算法与区域增长的方法进行了比较。实验证明,我们的算法相对区域增长有一定的优势。 Pavement detection plays an extremely important role in the autopilot system. Its specific application includes detection assistance, obstacle avoidance and automatic navigation. Vision-based pavement detection is the main classification of each pixel in the image, to distinguish whether it is the road. Most pavement detection algorithms so far have been applied during the day. In this article, we focus on the nighttime pavement detection. We use a near-infrared camera to capture nighttime images. In the detection, the plane reflection model is first used to fit the pavement part of the image. Then, a pixel-based classification method is used to classify each pixel in the image. In the experimental part, we compared our algorithm with the method of regional growth. Experiments show that our algorithm has some advantages over regional growth.