基于自适应图像分割自动提取道路中心线

The automatic road centerline extraction based on adaptive image segmentation

道路是城区地理空间信息中最重要的基础设施之一,从高分辨影像中自动、快速的提取道路特征,是快速更新城市道路网信息的重要途径。文中在分析道路基本特征的基础上,选择基于自适应结构元素的形态分析算法提取初始道路区域;引入面积和长宽比等形状指数,得到较精确的道路信息;最后,采用Hilditch细化算法,并进行优化处理。实验证明,该道路提取过程中无需人工设置参数,且能够得到具有较高完整性和正确性的道路中心线。

The road is one of the most important infrastructures in urban geospatial information. Automatic and fast road extraction from high resolution image is an important way of updating the urban road network fast. Thus a new automatic method is proposed to extract the road feature. First, the adaptive structure element is adopted in the image segmentation. And then, the precise road features can be extracted under the constrains of area and length-breadth ratio （referred to as L/B）. Final, the road line is obtained by using Hilditeh algorithm and optimizing process. The experiments indicate that the road extraction without artificial set parameters in the process can get the centerline of the road with high completeness and correctness.