一种基于卷积运算的遥感影像道路提取方法

Convolution-based Remote Sensing Image Road Extraction Method

根据遥感影像道路特征,提出一种基于卷积运算和数学形态学的道路提取方法。首先通过拉普拉斯算子对图像行锐化处理;再用卷积运算获得线特征加强系数,在加强道路线特征的同时,背景灰度恒定区域被弱化;然后去除短线和块状噪声,适当地加入人工干预进行道路修剪;最后用数学形态学细化处理得到道路网。实验结果表明,该方法能有地从遥感影像中提取出道路。

Path characteristics based on remote sensing images,a convolution-based computing and mathematical morphology methods of road extraction.First of all,through the Laplacian sharpening of image processing, convolution operation used to obtain line features of the strengthening factor in enhancing the characteristics of the road line at the same time,a constant gray background weakened region.Then removed the short-term and massive noise,due to join the road pruning manual intervention,and finally using mathematical morphology had been refined to deal with the road network.The experimental results show that this method can effectively extracted from remote sensing images of the road.