基于对数变换的NPP-VIIRS夜间灯光遥感影像在城市建成区提取中的应用

发布时间：2018-07-10 04:31

本文选题：对数变换 + NPP-VIIRS　；参考：《华东师范大学》2017年硕士论文

【摘要】：不论从区域尺度还是全球尺度而言,城市建成区范围的界定都是监测和管理城市发展的重要基础,也是研究社会政治、经济、文化、环境的重要条件。得益于科技的迅猛发展,城市扩张日益迅速,如何快速而准确地提取城市建成区成为研究热点。城市的起源、形成、规划、发展都与人类活动存在紧密联系。已有研究表明,夜间灯光遥感数据能够有效地反映人类活动程度,进而为城市建成区的提取提供了新的思路。至今为止,很少有基于Suomi National Polar-orbiting Partnership-Visible Infrared Imaging Radiometer Suite(NPP-VIIRS)夜间灯光遥感影像提取城市建成区的研究。本文基于NPP-VIIRS月合成夜间灯光遥感数据,系统地探究了城市建成区的提取方法。本文的研究内容和成果主要分成以下几个方面:1)探讨了基于NPP-VIIRS月合成夜间灯光遥感数据提取城市建成区的关键难点。研究通过系统地比较 Defense Meterological Satellite Program-Operational Linescan System(DMSP-OLS)稳定夜间灯光遥感数据与NPP-VIIRS月合成夜间灯光遥感数据的特性,分析两类数据的优缺点及其在空间分布上的异同点,探明了城市建成区、非城市建成区及过渡区域在两类夜间灯光遥感数据中的对应分布特征。研究表明,NPP-VIIRS月合成夜间灯光遥感数据的像元辐射值分布空间差异性大,数据辐射值差异也大,这不利于城市建成区的提取。2)解决了利用NPP-VIIRS月合成夜间灯光遥感数据提取城市建成区的关键难点。针对关键难点,提出了对数变换数据处理方法来解决。此方法可以有效地缩小夜间灯光数据的亮度差异,减小数据异方差,突出城市与非城市之间的分界区域。3)本研究比较了使用城市周长-突变检测法、基于拉普拉斯改进的Sobel边缘探测法、邻近统计分析法以及分水岭分割算法,基于NPP-VIIRS月合成夜间灯光遥感数据及对数变换的NPP-VIIRS月合成夜间灯光遥感数据提取的城市建成区精度。研究表明,对数变换处理能够有效提高基于NPP-VIIRS月合成夜间灯光遥感数据的城市建成区提取精度,处理之前和处理之后的平均Kappa系数分别为0.54和0.82。4)本研究根据实验结果,分析并比较了四种城市建成区提取方法的优缺点,并提出了对应的改善措施。研究结果表明,整体而言,城市建成区提取结果最好的是分水岭分割算法;最邻近统计分析方法识别的城市建成区形状特征与基于验证数据得到的城市建成区的形状特征最相似;突变检测法不适用于基于原始的以及对数变换的NPP-VIIRS月合成夜间灯光遥感数据的城市建成区提取的研究。
[Abstract]:In terms of regional scale and global scale, the definition of urban area scope is an important basis for monitoring and managing urban development, and is also an important condition for the study of social politics, economy, culture and environment. With the rapid development of science and technology, urban expansion is becoming more and more rapid. The origin, formation, planning and development of cities are closely related to human activities. Studies have shown that night lighting remote sensing data can effectively reflect the extent of human activity, and thus provide a new way of thinking for the extraction of urban built-up areas. Up to now, few researches have been done to extract urban built-up areas from night light remote sensing images based on Suomi National Polar-orbiting Partnership-Visible Infrastructure Imaging Radiometer Suite (NPP-VIIRS). Based on NPP-VIIRS month synthetic night light remote sensing data, the extraction method of urban built-up area is studied systematically in this paper. The research contents and achievements of this paper are divided into the following aspects: 1) the key difficulties in extracting urban built-up areas based on NPP-VIIRS month synthetic night lighting remote sensing data are discussed. By comparing the characteristics of the night light remote sensing data stabilized by the Defense Satellite Program-Operative Linescan system (DMSP-OLS) and the NPP-VIIRS synthesized night light remote sensing data systematically, the advantages and disadvantages of the two kinds of data and their differences and similarities in spatial distribution are analyzed. The corresponding distribution characteristics of non-urban built-up areas and transition areas in two types of night light remote sensing data. The results show that the spatial difference of pixel radiation value of NPP-VIIRS synthetic night light remote sensing data is great, and the difference of data radiation value is also great. This is not conducive to the extraction of urban built-up areas (.2) and solves the key difficulties in extracting urban built-up areas by using NPP-VIIRS month synthetic night lighting remote sensing data. In view of the key difficulties, a logarithmic transformation data processing method is proposed to solve the problem. This method can effectively reduce the luminance difference of night lighting data, reduce the heteroscedasticity of the data, and highlight the boundary area between city and non-city. Based on Laplacian improved Sobel edge detection method, proximity statistical analysis method and watershed segmentation algorithm, the precision of urban building area extraction based on NPP-VIIRS month synthetic night lighting remote sensing data and logarithmic transformation NPP-VIIRS month synthetic night lighting remote sensing data is obtained. The results show that the logarithmic transformation process can effectively improve the precision of urban building area extraction based on NPP-VIIRS monthly synthetic light remote sensing data. The average Kappa coefficients before and after processing are 0.54 and 0.82.4 respectively. This paper analyzes and compares the advantages and disadvantages of four methods for extracting urban built-up areas, and puts forward corresponding improvement measures. The results show that, on the whole, the watershed segmentation algorithm is the best one. The shape features of the urban built-up areas identified by the nearest neighbor statistical analysis method are most similar to those of the urban built-up areas based on the validation data. The mutation detection method is not suitable for the extraction of urban built-up areas based on the original and logarithmic transformation NPP-VIIRS month synthetic night light remote sensing data.
【学位授予单位】：华东师范大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：P237

【相似文献】