基于多源数据融合的冲沟参数提取研究

发布时间：2018-02-20 22:18

本文关键词： 冲沟侵蚀资源三号卫星影像融合冲沟提取　出处：《鲁东大学》2016年硕士论文　论文类型：学位论文

【摘要】：近年来,随着空间信息及测量技术的发展,特别是高分遥感、立体摄影测量、雷达干涉测量、RTK-GPS测量、激光三维扫描等新技术的涌现,为冲沟侵蚀的研究提供了大量的多源数据,丰富了冲沟侵蚀研究的方法和手段。但这些数据多单独应用于冲沟侵蚀研究,各种数据的结合应用只存在于独立的交叉验证,缺乏真正的融合。紧密融合多源数据,构建基于多源数据融合的冲沟参数提取方法成为目前迫切需要解决的问题。基于此,本文选择冲沟侵蚀广泛分布的庵里水库西缘流域为研究区域,探讨多源遥感数据的融合技术,建立基于多源数据融合的提取冲沟参数方法,以期拓宽冲沟侵蚀研究方法,提高冲沟研究精度。获得了以下结论:(1)比较三种全色与多光谱影像融合方法,HSV变换影像图像亮度过高、光谱损失严重,目视效果较差。PCA变换影像信息量较丰富,但清晰度最差,光谱损失程度较大,目视效果最差。GS光谱锐化影像目视效果最好,光谱保持能力好,信息量丰富及清晰度较高。GS光谱锐化融合方法是一种适宜资源三号卫星全色与多光谱影像融合的方法。(2)基于GS光谱锐化、HSV变换及PCA变换影像提取的冲沟数量均多于原始影像提取的冲沟数量,比基于原始多光谱影像解译的冲沟更准确。结合实测冲沟参数对基于融合影像提取的冲沟参数进行精度检验时发现,基于GS光谱锐化图像提取的冲沟参数精度最高,GS光谱锐化融合方法是一种适合冲沟参数提取的融合算法。(3)采用SRM乘法融合法和Gram-Schmidt光谱锐化方法、HSV变换三种方法对资源三号卫星多光谱图像进行DEM与多光谱影像融合,三种融合方法都将SRM数据综合得到原始图像中,实现了图像的正立体化,提高了遥感图像解译地形参数的能力;SRM乘法校正后图像光谱保持能力好,适宜定量遥感分析,并且对平坦地区地物光谱信息保持程度最好;HSV融合图像光谱失真明显,但信息丰富、清晰度好且地形正立体感强,适合地貌学研究;GS融合图像正立体感明显,光谱失真较明显,信息量与清晰度不如HSV融合法,适合通用制图应用。(4)选取HSV变换方法对DEM数据及多光谱数据进行了融合,融合结果一方面克服了原始图像表现出的地形视觉混淆,另一方面,赋予了图像丰富的三维地形信息,有利于对冲沟的判读;基于融合影像,以结合沟沿线、沟底线坡度阈值与目视解译的冲沟提取方式,提取实验区76条冲沟,冲沟总面积1.13km2,沟壑密度5.2km/km2,为极强烈侵蚀;以基于DOM影像提取的冲沟作为标准冲沟,对基于融合影像提取的冲沟参数进行精度检验,基于融合影像提取的冲沟有效沟沿线及沟底线比率均为91%,明显高于原始影像,基于融合影像提取的冲沟参数精度高于多光谱影像;多光谱影像与地形数据融合方法可以提高冲沟参数解译精度。
[Abstract]:In recent years, with the development of spatial information and measurement technology, especially high-score remote sensing, stereo photogrammetry, radar interferometry RTK-GPS measurement, laser three-dimensional scanning and other new technologies, a large number of multi-source data have been provided for the study of gully erosion. It enriches the methods and means of gully erosion research, but most of these data are used separately in gully erosion research, and the combination of all kinds of data only exists in independent cross-validation, lacking of true fusion. It has become an urgent problem to construct a method of extracting gully parameters based on multi-source data fusion. Based on this, this paper selects the western edge basin of Anli Reservoir, where gully erosion is widely distributed, as the research area. This paper discusses the fusion technology of multi-source remote sensing data and establishes a method of extracting gully parameters based on multi-source data fusion in order to broaden the research method of gully erosion. In order to improve the precision of gully research, the following conclusions are obtained: (1) comparing the three fusion methods of panchromatic and multispectral images, the HSV transform images are too bright, the spectral loss is serious, the visual effect is poor. The PCA transform image is rich in information, but the definition is the worst. The degree of spectral loss is large, the visual effect is the worst. GS spectral sharpening image has the best visual effect, and the spectral retention ability is good. The method of spectral sharpening fusion with rich information and high definition. GS is a suitable method for fusion of panchromatic and multispectral images of satellite No. 3.) based on GS spectral sharpening and PCA transform, the number of gullies extracted is many. The number of gullies extracted from the original image, It is more accurate than that based on the interpretation of the original multispectral image. GS spectral sharpening fusion method based on GS spectral sharpening image extraction is a fusion algorithm suitable for gully parameter extraction. It adopts SRM multiplication fusion method and Gram-Schmidt spectral sharpening method. The multispectral image of Resource-3 satellite is fused with DEM and multispectral image. All of the three fusion methods synthesize the SRM data into the original image, realize the orthotropic of the image, improve the ability of interpreting the terrain parameters of the remote sensing image and improve the spectral retention ability of the image after the correction by the SRM multiplication, which is suitable for quantitative remote sensing analysis. The spectral distortion of HSV fusion image is obvious, but the information is rich, the definition is good and the terrain is positive stereoscopic, which is suitable for geomorphology study of GS fusion image, and the spectral distortion is obvious. The amount and clarity of information is inferior to that of HSV fusion method, which is suitable for general cartographic application. (4) HSV transform method is selected to fuse DEM data and multispectral data. The fusion results not only overcome the terrain visual confusion of the original image, but also overcome the visual confusion of the original image on the other hand. The images are endowed with abundant 3D terrain information, which is beneficial to the interpretation of the gully, and based on the fusion image, 76 gullies are extracted from the experimental area by combining the gully baseline slope threshold with visual interpretation. The total area of the gully is 1.13km2, and the density of the gully is 5.2km2, which is very strong erosion. The gully extracted based on the DOM image is taken as the standard gully, and the precision of the gully parameters extracted from the fusion image is tested. The ratio of effective gully and furrow bottom line based on fusion image is 91R, which is obviously higher than that of original image, and the precision of gully parameter extraction based on fusion image is higher than that of multispectral image. Multispectral image and terrain data fusion method can improve the precision of gully parameter interpretation.
【学位授予单位】：鲁东大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：S157.1

【参考文献】