【作者】 胡娟；

【导师】 安裕伦； 杨柏林；

【作者基本信息】 贵州师范大学 ， 环境科学， 2006， 硕士

【摘要】 文章以覆盖贵州省29景CBERS-01遥感图像标准化处理及CBERS-01在喀斯特山区数字图像增强技术实验为研究重点,选择都匀市为典型样区,进行喀斯特山区土地利用动态监测应用研究,探索CBERS-01在喀斯特山区土地利用最理想的图像处理方法及分类方法。以“3S”技术为总的技术背景,以遥感数字图像处理技术为主要技术支持,以中巴地球资源01/02星为遥感数据源,运用专业遥感图像处理软件和地理信息系统软件ArcGIS8.3、ARCview开展工作。在遥感数字图像预处理中,通过多次试验,采用纹理滤波法消除条带噪声,以96年TM影像为参考标准进行二阶多项式几何精校正和最邻近像元法重采样,误差控制在1个像元内。镶嵌试验中,选择先单轨道多波段直方图匹配拼接,再进行轨道间直方图匹配拼结的方法,最终获得接边和色彩匹配基本满意的全贵州CBERS-01影像镶嵌图。在图像增强处理中,运用辐射增强、空间增强、图像运算、色彩增强、多源遥感数据融合等多种方法,经过多次试验对比,发现对线性拉伸后的图像进行非定向边缘增强,得到的纹理信息图像再与原始图像作叠加运算,得到的合成图像既突出了图像的细节信息又突出了纹理信息。介绍分析了各种植被指数,利用非线性比自动提取植被信息。比较CBERS-01图像K-L变换和K-T变换的两个主要成分,发现缨帽变换能较好地分离土壤和植被。基于RGB-IHS变换和主成分变换模式,将CBERS-01与ASTER、TM、SPOT等数据进行多光谱影像融合实验。另外,展开CBERS-01与TM在灰度信息、纹理信息、边缘信息及宏观特点等方面的对比研究。建立CBERS-01多种图像处理相关数字图像产品卫星影像库10余个。最后以都匀市为例,以CBERS-01/02为数据源研究喀斯特山区土地利用监测中水体与阴影混分以及耕地与建设用地混分的问题,建立了利用波段相关性自动提取水体信息和基于光谱分析的耕地信息自动提取模型,实现了CBERS替代价格昂贵的国外遥感影像TM、SPOT,对喀斯特地区进行大比例尺土地资源的动态监测。更多还原

【Abstract】 With the standard processing of 29 remote sensing images of Guizhou province from CBERS-01 and the experiment on CBERS-01 in digital image enhancement technique of karst mountainous areas as the studying emphasis, the paper chooses Duyun city as the typical region to study dynamic monitoring of land use change in karst areas and discuss the best image processing and sorting methods of land-use with CBERS-01.With 3S (GIS, GPS and RS) techniques as the background and remote sensing digital image processing technique as the support, CBERS-01/02 as the remote sensing data resource, this paper carries out the study with such professional software as ArcGIS8.3 and ARCview .In the course of remote sensing digital image preparation, the paper, based on many experiments, chooses histogram match for radiation correction, texture filter method for strip noise reduction, and, with TM image in 1996 as the standard, carries out geometric correction and re-sampling of multi-band combined colored image by means of binary multinomial and closest pixel, with the error of one pixel. In the mosaic experiment of CBERS-01, this paper chooses the method of one-track multi-band histogram mosaic plus inter-track histogram mosaic to decrease the color difference of images and improve the quality of images.In course of image enhancement processing, this paper, by the means of radiation enhancement, spatial enhancement, image operation, color enhancement and RS data merge, thinks that the combined image by textural information image and original image can reflect both detail information and textural information while carrying out non-directional edge of the linear stretch image. Then the paper introduces vegetation indexes and extracts automatically the vegetation information through non-linear ratio. By comparison of K-L change and K-T change, this paper thinks tasseled cap change can separate vegetation from soil better. In addition, on the basis of RGB-IHS and principle component analysis, this paper carries out a multi-spectral image merge experiment of CBERS-01 with ASTER, TM and SPOT, then makes a comparison on CBERS-01 and TM from the information of brightness, texture and edge, and macro characteristics. Finally this paper establishes ten image databases including non-linear ratio, non-linear stretch, ratio method, difference method, vegetation indexes (NDVI, RVI, DVI, PVI, SAVI) and spatial transform of principle component.The paper, with a special reference to Duyun city and on the basis of更多还原