多源数据协同的城市道路提取

城市道路的高精度提取可为城市三维表达、城市地形分析、城市建设规划、交通导航等提供数据基础和支撑。本文以合肥市局部城区为试验区，以开源路网、街景图像和遥感影像为数据源，在利用最大似然法进行初提取的基础上，通过空间分析、统计分析、几何量测、最小二乘拟合等方法进行粘连分割、缺失处理和交叉口细化等关键处理，构建了多源数据协同的城市道路提取方法，并对提取结果进行了精度评价和分析。试验结果表明，本文提出的城市道路提取方法优于最大似然和面向对象方法，提取总体精度为96.65%，Kappa系数为93.71%，道路宽度偏离标准差为0.03m，特别是对同物异谱、同谱异物及遮挡等造成的信息提取不全问题具有良好的改善效果。

Urban road extraction based on multi-source data

The high-accuracy extraction of urban roads can provide data basis and support for many fields, for instance, three-dimensional urban expression, urban terrain analysis, urban construction planning and traffic navigation. Comprehensively combining the advantages of open source road network, street-view images and remote sensing images, and taking the part of Hefei city as the experimental area, an extraction approach for urban road is proposed. Firstly, the maximum likelihood method is used to extract the urban road. Secondly, the void is filled and the connecting areas that does not belong to the urban road are divided by using the open source road network and street view data. Thirdly, the missing parts caused by cover, like vegetation, are disposed of via the width of each road which measured by street view. Finally, the intersections of urban road are improved. Spatial analysis, statistical analysis, geometric measurement, least square fitting methods are involved in the process. To assess the performance of the proposed method, the precision of urban road extraction is evaluated and analyzed. The experimental results indicate that the urban road extraction method proposed in this paper can extract urban road with high precision and the extraction precision is 96.65%, Kappa coefficient is 93.71% and the standard deviation of road width is 0.03 m. In particular, it can improve the incomplete information extraction caused by different spectrum of the same object, different object with the same spectrum and the problem of cover.