基于高分遥感影像的断裂道路连接方法

Fracture road connection method based on high score remote sensing image

摘要:

针对高分遥感影像道路中心线提取过程中出现的“毛刺”与“断裂”道路问题，提出一种去除“毛刺”和道路中心线连接的方法。首先利用SVM得到初始分类图，然后通过边缘滤波、纹理滤波和形状滤波去除非道路区域，得到道路提取图。接着利用细化算法得到初始道路中心线，本文提出一种新的“交点”搜索规则，使用基于交点的“毛刺”去除算法对道路中心线中的“毛刺”进行去除。最后提出一种基于“待连接点”的连接方法对“断裂道路”进行连接，得到光滑和完整的道路中心线提取图。实验表明，“毛刺”现象去除效果明显，“断裂道路”连接完整，可以有效的提取路网信息。

Abstract:

Aiming at the problem of “burr” and “fracture” roads in the process of extracting road centerline of high score remote sensing image, a method to remove the connection between “burr” and road centerline is proposed. First use SVM to get the initial classification map, and then remove the non-road areas through edge filtering, texture filtering and shape filtering to get the road extraction map. Then use the refinement algorithm to get the initial road centerline. This paper proposes a new "intersection point" search rule, which uses the intersection-based "glitch" removal algorithm to remove the "burr" in the road centerline. Finally, a connection method based on the “point to be connected” is proposed to connect the “broken roads”, and a smooth and complete road center line extraction map is obtained. Experiments show that the "burr" phenomenon has obvious removal effect, and the "fracture road" is completely connected, which can effectively extract road network information.