基于OSM数据和遥感影像的道路正样本制作方法

Method for creating road positive samples based on OSM data and remote sensing images

为深入探讨无先验样本情况下道路正样本的制作,以OSM数据为辅助,提出了一种道路正样本制作方法,并与传统人工制作方法进行对比分析。结果表明,文中方法的提取完整度为94.21%,正确率为97.02%、提取质量达到91.56%,该方法可有效地解决阴影、树木遮挡等情况下,自动制作多区域、不同类型的道路正样本,且有良好的提取质量。

This paper aim s to explore the production of road positive samples deeply without prior samples.The study proposes a method of road positive sample production assisted by OSM data,and compares and analyzes with the traditional manual production method.The results show that,with the proposes method,the extraction integrity is 94.21%,the accuracy is 97.02%,and the extraction quality is 91.56%.This method can effectively solve the problem of shadow,tree occlusion and other cases,and automatically produce positive samples of multiple regions and different types of roads,and has good extraction quality.