基于形态学阴影指数的高分二号影像建筑物高度估计  

Building Height Estimation from GF-2 Image based on Morphological Shadow Index

作　　者：田峰[1,2,3] 陈冬花 黄新利 李虎 姚国慧 赵俊鹏[2,3,6] 丁锋 邢菲 

机构地区：[1]新疆大学资源与环境科学学院,新疆乌鲁木齐830000 [2]新疆维吾尔自治区卫星应用中心,新疆乌鲁木齐830000 [3]乌鲁木齐北辰众星空间信息科技有限公司,新疆乌鲁木齐830000 [4]滁州学院,安徽滁州239000 [5]南京大学地理与海洋科学学院,江苏南京210000 [6]新疆农业大学林业研究所,新疆乌鲁木齐830000 [7]新疆农业大学草业与环境科学学院,新疆乌鲁木齐830000

出　　处：《遥感技术与应用》2017年第5期844-850,共7页Remote Sensing Technology and Application

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)

摘　　要：高分二号卫星是我国自主研发的高分辨率对地观测卫星,分辨率达到亚米级。利用高分二号卫星遥感影像,结合数学形态学和面向对象分类的思想进行城市建筑物高度估计。首先,利用多尺度分割将影像分割成对象;进而结合光谱、形状、形态学阴影指数(MSI)等特征面向对象进行分类,相对准确提取出建筑物的阴影并计算阴影的长度;最后,结合影像成像时的卫星、太阳和建筑物的几何关系模型进行建筑物高度估计,并利用实地测量数据进行精度评价和误差分析。结果显示,90%的估计结果绝对误差小于1m,说明该方法可以有效地从高分二号影像中提取建筑物高度,展现了国产高分辨率遥感影像提取城市建筑物信息的巨大潜能。GF-2 is a high resolution earth observing satellite with sub-meter resolution which is developed by our own technique.To estimate urban building height based on GF-2 remote sensing image combined with the idea of mathematical morphology and object-oriented classification.First of all,segment image based on multi-scale segmentation.Then extract shadow and calculate its length based on object-oriented classification combined with spectral,shape,Morphological Shadow Index（MSI）and other features.In the end,estimate building height based on the geometrical model of satellite,sun and building and then accuracy evaluation and error analysis are carried out by using the field measurement data.Experimental results showed that 90% of the buildings＇ absolute error is less than 1 m.This experiment demonstrate that the method can extract the height of urban building from the GF-2 image effectively and the immense potential of domestic high resolution remote sensing image in applications on urban building information extraction.

关 键 词：高分二号 形态学阴影指数(MSI) 面向对象 阴影提取 建筑物高度 

分 类 号：TP79[自动化与计算机技术—检测技术与自动化装置]