基于LandSat8 OLI数据的山区阴影信息检测与提取  

Detection and Extraction of Mountain Shadow Information from LandSat8 OLI data

作　　者：池毓锋 赖日文[1] 闫琦[1] 余莉莉[1] 苏艳琴[1] CHI Yufeng LAI Riwen Yan Qi YU Lili SU Yanqin(College of Forestry, Fujian Agriculture and Forestry University, Fuzhou 350002, Chin)

机构地区：[1]福建农林大学林学院,福州350002

出　　处：《山地学报》2017年第4期580-589,共10页Mountain Research

基　　金：生态林种科研基地建设工程项目(61201400814);森林持续经营研究(ky0180081)~~

摘　　要：为快速获取山区遥感影像上的阴影干扰区域,探究一种简便、高效、精确的遥感影像中阴影干扰区域的方法,具有重要的意义。以福建省长汀县为研究区域,搜集2016年3月美国陆地卫星影像数据与ASTER影像计算的GDEM V2产品。基于土地利用分类体系测量的6类地面物体光谱信息与影像波段信息,优选影像的光谱波段并重新组合;选取Sinh函数与Max函数建立c!算法,对LandSat8 OLI影像进行差异化计算,通过二分式判别规则初步提取阴影区域;加入由数字高程模型计算的坡度信息,剔除水域与坡度较为平缓的地物等干扰信息,精确提取山区阴影区域。设立网格随机设置精度验证点验证精度,最终总体精度达到99.06%,Kappa系数为0.98。结果表明,实验方法对于LandSat8 OLI影像提取阴影可行性高,检测效果与提取结果较c3算法与SVI指数更好。Shadows in remote sensing images contain cues about the shape and the relative position of objects,as well as the characteristics of surfaces and light sources. For this reason,the problem of shadow detection had been increasingly addressed over the past years. In this study,a new extraction method（ c!） was established to extract the shadow information in mountainous region from satellite image data of Changting County,Fujian Province. The c! algorithm was established based on the free LandSat8 Operational Land Imager（ OLI） image taken on 22 March,2016 and the product of GDEM V2 image taken by ASTER satellite. This method was applied to differentiate shadows and non-shaded areas. The c! algorithm included band selection,band recombination,ground feature spectrum analysis,function construction,slope factor extraction and other important portions. In the part of function construction,function Sinh and function Max were used to calculate the image band variation. The 8 bit image value was confirmed through the threshold segmentation,that is,199. Then the slope factor（ with a value of18 degree） extracted from GDEM V2 image was also combined within the algorithm. Finally,the mountain area shadow image and some other sample results were outputted using c! method and c3 method in the same research area,respectively. According to the grid random verification point data,the extraction results were verified and then compared with the results from c3 method. It showed that c! method in combination with spectrum and some other factors can significantly improve the extraction accuracy and effect. The overall accuracy and Kappa coefficient were 99. 06% and 0. 98,respectively. The study results suggested that the experimental method had high feasibility for shadow information detection and extraction on LandSat8 image. Moreover,it was simpler and more excellent when compared with SVI index method or c3 algorithm.

关 键 词：LandSat8 OLI影像 c!算法 阴影 坡度因子 

分 类 号：TP79[自动化与计算机技术—检测技术与自动化装置] P951[自动化与计算机技术—控制科学与工程]