摘要： 针对机场在遥感图像中成像的复杂性与多样性，提出了一种新的机场跑道的自动识别方法。该方法将跑道识别分为定位和识别两个步骤：首先在定位步骤中基于变形模板模型理论，建立了跑道的平行线模型及似然函数，并采用Meropolis 算法确定跑道的准确边界；其次，通过分析跑道的标志特征和结构特征确立了跑道识别的通用性判别规则。通过对Spot 图像的实验证明，该方法能够实现复杂环境下的跑道自动识别。

关键词: 计算机图像处理；遥感图像；跑道；变形模板

Abstract: A novel recognition approach of runway is proposed aimed at the complexity and variety of airport images. First, a parallel edges model and corresponding likelihood function are proposed and Metropolis algorithm is used to accurately localize the edge of runways. Furthermore,effectively identification rules which analyze both structural and mark features of runways are used in this approach. The results of experiments on spot satellite images show that this approach can successfully accomplish runway recognition tasks even in complex environment

Key words: Computer image processing; Remote sensing image; Runway; Deformable template