基于改进SUSAN算子的车道线检测算法

Lane detection method based on improved SUSAN operator

为了准确识别道路的车道线,采用45°Sobel边缘算子对中值滤波后的道路图像进行增强,进而采用SUSAN算子和Otsu算法相结合的方法将图像分割,利用分区Hough变换进行拟合,识别出车道线.实验结果表明,采用改进的SUSAN算子分割后的道路图像能够准确提取车道线参数,拟合出车道线,去除噪声,节省了后续Hough变换的时间,提高了算法的抗噪性能和实时性.

In order to identify the lane of the road image accurately, the road image which was filtered by median filter was en-hanced by 45° Sobel algorithm. Then the image was segmented by using the method of combining the SUSAN operator and the Otsu al-gorithm. At last, the lanes were identified by zoning Hough transformation. The results show that the road image which was segmented by improved SUSAN operators can obtain the effective feature points, remove noise, and save time of Hough transformation, improve the anti-noise performance and real-time performance of algorithm.