×
【作者】 胡兴堂
【导师】 童庆禧;郑兰芬;张兵
【 学位年度 】 2006
【论文级别】 博士
【关键词】 高光谱遥感;光谱代数表达;水质监测;影像压缩、分形特征编码;平台构建 ;;;
【Key words】 hypersepctral remote sensing;water parameter extraction &monitoring;hyperspectral algebra structure analysis & presentation;image compression;system integration.
【中文摘要】
本文从系统工程的角度出发,结合高光谱水质遥感的特点与环保行业需求,从水质环境遥感业务和软件工程两方面出发进行研究,讨论了高光谱遥感数据的多分辨率表达模型,海量高光谱数据的快速处理与访问技术,海量高光谱遥感数据的光谱保形高效压缩技术,针对海量多源多平台(航空/航天)遥感数据进行环保业务反演系统构建。本文系统研究了遥感监测平台从建立到运行包括数据获取预处理,数据分析,专题产品反演,产品发布的整体架构,产品定义和业务规范的建立,提供了针对水环境遥感监测的可行业务化平台涉及到的关键技术。从高光谱数据的自身特点和光谱数据库以及图谱合一的固有特点出发,充分考虑航空高光谱遥感特殊性,构建面向应用的高光谱数据处理与分析平台。并以次为基础结合水质遥感业务构建了业务化的水环境监测系统REMS。同时,以江苏太湖为示范区,针对环境遥感监测业务化运行要求,构建了环境遥感监测软件系统(REMS V1.0),基本形成多类型、多专题、多空间分辨率、多光谱分辨率、多时相环境遥感数据的综合处理和应用能力。
结合作者在过去几年中参与构建高光谱图像处理与分析系统和我国环境遥感监测平台的实践和体会,在分析总结国内外大量相关文献和系统的基础上,本文以高光谱水质遥感监测系统构建为框架,以作者在参与系统建设过程中提出发展并实现的理论和算法为主线,组织整理这篇博士论文。本文的工作主要体现在三个方面:
①数据处理软件系统与遥感反演模型集成交叉:针对高光谱数据的空间特性、光谱特性以及海量数据特性,系统地给了出海量高光谱数据处理思路和解决方案。
②高光谱遥感处理系统与水质遥感监测应用交叉:集成了水色大气校正模型和生物光学反演模型,建立了适合我国国情的内陆水色遥感技术体系和业务流程。
③以江苏太湖流域水环境监测构建应用实例,建立了国内首个环境遥感监测系统,为环保行业进一步应用做铺垫。
作者对以下几个问题做了较为深入的研究:
(1)提出高光谱图像代数结构用于描述高光谱数据,总结提炼出高光谱数据处理的五大典型操作。提出光谱代数和波段代数进行光谱参量化。
(2)检验了高光谱数据水质参量反演模型并总结了数据处理流程,给出水色反演遥感数据的标准化处理流程,给出了叶绿素反演光谱波段选择与光谱波段组合设计,建立了流域典型水质污染反演模型,对六个水质参量(悬浮物、叶绿素a、高锰酸盐指数CODMn、总磷、总氮和透明度)的时空变换规律进行了检验和验证。同时给出了6S模型支持下的水色大气校正方法。在此基础上,拓展研究区域环境监测平台构建关键技术,建立流域水环境质量及典型生态要素可遥感监测指标体系。
(3)提出了海量高光谱数据谱形保真压缩方法,给出了基于光谱保形海量高光谱影像数据分形编码框架,同时提出了基于光谱自相似性数据压缩算法。本文的压缩算法基于光谱特征空间压缩,在压缩性能方面具有很好的优越性。
(4)研究了海量高光谱遥感数据处理平台构建涉及到的核心技术,包括海量高光谱图像显示技术、海量高光谱图像快速处理技术、海量高光谱图像快速存取技术、高光谱数据库构建技术。提出了高光谱图像数据结构,并构建了高光谱遥感影像分块金字塔模型,给出了针对高光谱海量影像数据快速金字塔算法。设计了高光谱图像处理系统架构并给出了具体实现。
(5)给出了高光谱水质遥感监测系统构建架构,提出了多层次的遥感数据关系模型,讨论了基于任务驱动太湖流域水质反演模式业务模式,给出了多摸式REMS视窗设计实例,讨论了分布式海量高光谱影像数据并行提取机制,设计并实现了高光谱水质遥感监测系统构组件模块。最后我们给出了太湖水质监测产品查询与发布系统REMSIS架构和实例。
【Abstract】
Design and Implement application software platform is an important feature of thesystem project of the Earth Observing Plan. In this thesis, we focus our research on Hyperspectral Remote Sensing Environment Monitoring System and its applications to water resources. To meet the requirements for water quality monitoring in China, a Remote-sensing Environmental Monitoring System (REMS) is introduced. REMS isthe first integrated system in developed for multi-resource, multi-temporal, and multi-thematic data processing and data analysis, and for distributing products for the monitoring of inland water pollution using remote sensing technology. REMS provides the ability to quickly extract the major characteristics of water resources, such as chlorophyll content, total suspended matter (TSM), yellow substance, and blue algae distribution. In order to improve the precision of water parameter extraction, new algorithms and functions are also developed and integrated into the REMS software platform. We select the Taihu Lake in Jiangsu Province, China, as the research area. Finally, we discuss some field applications constructed based on REMS.
My dissertation includes the following:
(1) The introduction of hyper-spectral data algebra structure to hyperspectral algebra structure analysis (HASA).
(2) The establishment of semi-empirical and analytical models to extract the major characteristics of water resources such as chlorophyll content, total suspended matter (TSM), Yellow Substance, and CDOM. The analytical models are based on bio-optical models and radiation-transfer theory, where the optical properties and water quality parameters have distinct physical meaning and universal applicability.
(3) The introduction of fractal-based image compression algorithms using wavelet transformation for hyperspectral images in Chapter four. This algorithm is superior to other traditional compression methods because it has high compression ratios, good image fidelity, and requires less computation time. I present a fast fractal image coding methodology based on wavelet decomposition. The subimage blocks are improved by Pyramidal Haar transform. Furthermore, the improvement of the Quadtree partition is also discussed. My simulations show that the coding time is 100 times faster when the same coding result is retained. The HV and Quadtree partitioning and the domain-range matching algorithms have also been improved to accelerate the encode/decode efficiency.
(4) The design of the most important components of the Hyperspectral Remote Sensing Image Processing and Analysis System, including tools for input/output, preprocessing, data visualization, information extraction, conventional image analysis, advanced tools, and integrated interface to connect with general spectral databases. The base architecture was specially designed and implemented to meet the requirements for the rapid preprocessing of imaging spectrometer data and the easy prototyping of algorithms. Some new methodologies for data analysis and processing were developed and applied to obtain results based on the architecture including mineral identification, agriculture investigation, and urban mapping. The key technologies of the Hyperspectral Remote Sensing Image Processing and Analysis System include the following: image tiling and processing, execution architecture, general data preprocessing, hyperspectral data calibration, hyperspectral data analysis, and image classification.
(5) The design of a new architecture for REMS. REMS provides the ability to extract the major characteristics of water resources, such as chlorophyll content, total suspended matter (TSM), and yellow substance. The most important components of REMS are presented in chapter Six, including tools for multi-resource data input/output, preprocessing, data visualization and mapping, environment information extraction, conventional image analysis, advanced tools for eco-environment modeling, and integrated interface to connect with general spatial and spectral database and water quality monitoring database. We established a REMS internet information publishing system REMS IS. Integrated with WebGIS and RS technology, REMS IS has four tiers and five subsystems, so that it can meet the specification of the B/S and C/S architecture. The four-tier architecture includes the following: the REMS application server, the middle data server, the web server, and the client. The REMS application server is composed with two important modules: the atmosphere correcting module and the water quality parameter inversion module 。。。。。。。。。...