高分影像建筑物阴影优化提取与高度估算

Shadow optimization extraction and height estimation on high resolution imagery

针对现有方法提取阴影效率慢,提取不完整,估算过程未能实现半自动化甚至自动化的问题,该文基于高分二号影像,提出一种将K-means图像分割算法与阴影后处理结合一体应用在建筑物阴影提取的方法:首先,选择建筑物间隔稀疏,结构规则的城郊区域,利用K-means图像分割获取建筑物阴影、建筑物2类以提取建筑物阴影;其次,通过形态学算法、Canny边缘检测等对阴影后期处理,去除小区域及孔洞填充,边缘信息检测,获取最终建筑物阴影;最后,根据太阳、卫星、建筑物以及阴影长度之间几何关系计算建筑物高度。考虑研究区域户型,每层楼高以2.8 m量测建筑物实际高度作为验证,实验结果表明:利用K-means图像分割能有效提取出阴影区域,与后期阴影优化策略结合,大幅度改善了阴影区域的完整性,获取建筑物高度信息自动化程度得到提高。

In view of existing methods for extracted shadows are slow,extracted incompletely,and the estimation process fails to achieve semi-automation or even automation.In this paper,based on the GF-2 imagery,we would combine the K-means image segmentation algorithm with the post-processing and then applied to the building shadow extraction.Firstly,we chose the buildings with sparse building spacing and structural rules in suburban area.The K-means image segmentation that obtained building shadows and buildings sample,extracted building shadows.Secondly,the shadow were processed by morphological algorithm and Canny edge detection that removed small areas and filled hole,detected information of the shadow edge.And finally,according to the sun,the satellite and the geometric relationship between the buildings and shadow length,calculated the height of the buildings.Considering the type of the regions,the actual height of the building was measured by 2.8 m per floor.And then the experiment results showed that the K-means image segmentation could effectively extract the shadow area,which was combined with the late shadow optimization strategy could greatly improve the integrity of the shadow area,and enhance the automation degree of obtained building height information.