提出了一种新的从航空影像上提取道路中心线的半自动方法。该方法利用道路地物的几何特征和辐射特征,根据用户提供的少量道路种子点,利用自适应灰度模板对道路中心线上的每个点进行调整,最后采用最小二乘算法和牛顿迭代法对调整后的道路中心线上的每个点进行拟合,以此获得趋近于道路中心线的最佳折线。试验结果证明,该算法简单易行、效率较高,具有较好的鲁棒性和实用性。 A new semi-automatic method of extracting the center line of road from aerial image is proposed. The method uses the geometric features and radiation characteristics of the road features, adjusts each point on the road centerline according to a small number of road seed points provided by users, and adjusts each point on the road center line by using an adaptive grayscale template. Finally, the least squares algorithm and Newton’s iterative method Fit each point on the adjusted road centerline to get the best polyline that approaches the centerline of the road. Experimental results show that the algorithm is simple, efficient and has good robustness and practicability.