联合星载光学与SAR图像的城市大面积建筑物高度快速提取

发布时间：2018-03-20 18:08

本文选题：星载高分辨率图像　切入点：建筑物高度　出处：《测绘学报》2017年07期 　论文类型：期刊论文

【摘要】：星载高分辨率光学图像与SAR图像广泛应用于城市建筑物高度提取,但光学图像存在缺少相关卫星参数的情况,而SAR图像则存在散射特征不完整以及提取效率低等缺陷。针对以上问题,本文提出一种联合高分辨率星载光学与SAR图像的城市大面积建筑物高度快速提取方法。首先,结合支持向量机(SVM)和形态学阴影指数(MSI)快速提取光学图像中的阴影并自动测量阴影长度;之后选择多个合适样本,基于模型匹配法从SAR图像中提取高度;最后将高度与阴影长度作线性回归分析,建立数学模型来提取其他建筑物的高度。该方法将不同卫星系统的数据和特征相结合,互相弥补各自缺陷,不仅提高了效率、降低了成本,同时满足精度要求。
[Abstract]:Spaceborne high-resolution optical images and SAR images are widely used to extract the height of urban buildings, but the optical images are lack of satellite parameters. However, SAR images have some defects, such as incomplete scattering characteristics and low extraction efficiency. In view of the above problems, this paper proposes a fast method for extracting the height of large areas of buildings in cities by combining high-resolution space-borne optics with SAR images. Combining support vector machine (SVM) and morphological shadow index (MMSI) to extract shadows from optical images and measure shadow length automatically, then select several suitable samples and extract height from SAR images based on model matching method. Finally, the height and shadow length are linear regressed, and the mathematical model is established to extract the height of other buildings. This method combines the data and features of different satellite systems, and makes up for each other's defects, which not only improves the efficiency, but also improves the efficiency. The cost is reduced and the precision requirement is met at the same time.
【作者单位】：新疆大学资源与环境科学学院;新疆维吾尔自治区卫星应用中心;滁州学院;
【基金】：高分辨率对地观测系统重大专项(95-Y40B02-9001-13/15) 新疆青年科技创新人才培养工程(2013701035) 安徽省高等学校自然科学研究项目(KJ2016A526) 滁州学院科研启动基金项目(2014QD029)~~
【分类号】：P237

【相似文献】