基于DSM与阴影的彩色遥感影像建筑物提取

Building extraction from color remote sensing imagery based on DSM and shadows

针对大面积树木阴影的存在,使得利用阴影提取建筑物的结果与实际建筑物数量存在较大差距的问题,结合建筑物DSM数据,提出了一种基于DSM与阴影的彩色遥感影像建筑物提取方法.通过设定植被分割阈值剔除植被信息的影响,采用mean-shift算法对图像进行过分割,提取建筑物阴影区域,结合DSM数据处理,得到实际高度的NDSM图像,再次提取建筑物进行最终优化.实验结果表明,使用该方法提取建筑物比只根据阴影提取建筑物精度有所提高、形状较为规整,更接近实际地物.

In order to solve the problem that there was big gap between result of extracting the building based on shadow and real building number owing to the big area of trees shadow,combining with building DSM data,a new method of building extraction from remote sensing image with colors that based on DSM and shadow was presented. It eliminated the influence of vegetation information by setting the vegetation segmentation threshold. It extracted the shadow area of building from image partitioned by the mean-shift algorithm. It got the NDSM image that included actual height by processing of DSM. Then it extracted the building again and optimized extraction. The result indicated that this method improved precision to some extent than only according to the shadow for building extraction and the experiment configuration was more regular and closer to real surface features.