论文部分内容阅读
针对高分辨率遥感影像中道路交叉口难以用固定几何和光谱特征描述的问题,该文提出了一种基于像元结构指数的交叉口提取方法。采用自适应光谱异质性阈值来提取像元形状特征,较好地适应了不同交叉口之间的场景特征差异;构建了像元形状与交叉口结构的量化映射关系,赋予特征语义信息,从而保证了提取结果的合理性。基于城区复杂场景高分遥感影像的交叉口提取实验表明:所提方法能够准确定位交叉口中心点,并检测道路交叉结构;另外,该方法对于交叉口支路发生变化的情况也具有适用性,定性与定量分析表明了该文方法的有效性。
In view of the difficulty of using the fixed geometry and spectral features to describe road intersections in high resolution remote sensing images, this paper proposes a method of intersection extraction based on pixel structure index. The adaptive spectral heterogeneity threshold is used to extract pixel shape features, which is well adapted to scene feature differences between different intersections. A quantitative mapping relationship between pixel shape and intersection structure is constructed, which gives the feature semantic information Ensure the reasonableness of the extraction result. Experimental results show that the proposed method can accurately locate the center of the intersection and detect the intersection of the road. In addition, this method is also applicable to the case of the change of the branch of the intersection, Qualitative and quantitative analysis shows the effectiveness of the method.