一种基于数学形态学的遥感影像公路主干提取算法

An algorithm of main road detection from remote sensing image based on mathematic morphology

从空间遥感图像的复杂背景中提取较为细微的目标一直被视为重要而复杂的课题。空间遥感图像具有灰度丰富,地理形态含信息量大,背景噪声复杂不规则等特点。对于被夹在现代城市各类复杂建筑中间的公路而言,它们的提取和走向位置等信息的掌握具有普遍的实际意义和深远的研究价值。针对以上情况,提出一种从遥感图像中提取公路主干的基于数学形态学的算法。利用形态学中的计算手段,Top-hat算子,阈值分割以及维纳变换等方式对目标图像中的公路进行提取,实验结果证明了方法的有效性。该算法同时具有理论基础完备、算法简单、运算速度快以及抗噪能力强等优点。

Extracting the weak target from the remote sensing image which possesses complex background always occupies a significant and sophisticated subject, The remote sensing images are always characterized by varied histogram and complex noises. Speaking of the roads which are trapped within complex urban architecture, their extraction and direction information have practical meanings and profound researching values. Facing such condition, this paper put forward an algorithm of main road detection from remote sensing image which based on mathematic morphology. The basic compute of morphology, Top-Hat algorithm, threshold segmentation, Wiener transform and so on are applied here for extracting the main road target. The extracting result has shown the validity of the algorithm. Simultaneously this algorithm is also proved as theoretically mature, simple, efficient and strong in noise-resistance.