基于模糊连接度的卫星图像道路提取新方法

论文提出一种将相对模糊连接度和道路特征相结合进行卫星图像主干道路半自动提取的新方法。首先将原始图像划分为多个目标,对每一类目标分别人工选取种子点,然后计算出图像中各像素点与每一类种子点之间的模糊连接度,将其归属于模糊连接度值最大的一类,从而分离出道路目标。接着结合道路局部特征运用模糊连接度连接断裂路段,最后根据道路走向进一步完善整个道路轮廓。SPOT图像道路提取实验表明,新方法可以取得很好的效果。

Relative-Fuzzy-Connectedness-Based Road Extraction New Method from Remote Sensing Image

In this paper,the new method of road semi-automatic extraction based on relative fuzzy connectedness combined with road feature is proposed.The method firstly divides original image into multi-object,and chooses seed point for every object,and then calculates fuzzy connectedness between each pixel element and seed point of every object,and attaches the pixel element to that object which has the highest strength of connectedness,and then separates road object from multi-object.Secondly the method combines with the road characteristic and uses fuzzy connectedness to join the break road segment,and at last farther process to form the whole road network.Experiment indicates that new method may gain better results in SPOT image.