机载推帚式高光谱图像预处理技术研究与实现

发布时间：2018-02-03 19:40

本文关键词： 推帚式高光谱预处理多线程 OpenMP 辐射校正几何校正预处理系统　出处：《山东大学》2014年硕士论文　论文类型：学位论文

【摘要】：高光谱遥感的特点有更多波段数、光谱分辨率高、图谱合一、数据量大和信息冗余多等,随着高光谱遥感光谱分辨率的提高,可获得近似连续的地物光谱信息,对地表覆盖的识别能力更强。高光谱遥感图像预处理技术在高光谱图像处理与分析中占据重要的地位,是高光谱图像应用和分析的基础,直接影响后期的定量分析和信息的挖掘。所以,研究高光谱遥感图像预处理技术及其应用具有重要的意义。本文研究高光谱遥感图像预处理技术,并以此为基础实现高光谱遥感图像预处理系统软件。论文的主要工作如下： (1)介绍高光谱遥感图像预处理技术,包括常用的影像数据格式、光谱和辐射定标技术、高光谱图像辐射校正的具体实现过程、短波红外背景辐射消除方法和几何粗校正技术。重点分析基于位置姿态数据的几何粗校正技术,其中包括计算外方位元素、空间坐标转换和重采样方法等。 (2)为提高高光谱遥感图像预处理效率,研究几何校正并行处理技术。针对windows多线程技术OpenMP共享内存模型和MPI消息传递模型三种并行方式,深入研究几何校正并行处理的原理,并在计算机上编程实现,给出并行效率的比较。前两者应用于单机多核多线程处理,后者用于实现多机集群系统,重点研究OpenMP在提高数据处理和CPU效率方面的优势。 (3)在Windows环境下,基于VC++编程环境,使用微软基础类库(MFC),将算法的实现进行封装,完成高光谱遥感图像预处理软件的研制。依据软件设计原则构建软件的体系结构,设计实现软件系统的功能模块,如数据预整理模块、辐射校正模块、几何校正模块和快速处理模块等。几何校正模块采用多线程技术,快速处理模块提供一键处理和批量处理功能,保证数据处理的效率。
[Abstract]:Hyperspectral remote sensing is characterized by more bands, high spectral resolution, unified spectrum, large amount of data and redundant information, etc. With the increase of spectral resolution of hyperspectral remote sensing, approximately continuous spectral information of ground objects can be obtained. Hyperspectral remote sensing image preprocessing plays an important role in hyperspectral image processing and analysis, which is the basis of hyperspectral image application and analysis. Therefore, it is of great significance to study the preprocessing technology of hyperspectral remote sensing image and its application. In this paper, the preprocessing technology of hyperspectral remote sensing image is studied, and the software of hyperspectral remote sensing image preprocessing system is realized. The main work of this paper is as follows: This paper introduces the preprocessing technology of hyperspectral remote sensing image, including the commonly used image data format, spectral and radiometric calibration technology, and the realization process of radiation correction of hyperspectral image. The methods of eliminating infrared background radiation and rough geometric correction are analyzed, and the geometric coarse correction based on position and attitude data is analyzed, including calculation of external azimuth elements, spatial coordinate conversion and resampling. In order to improve the preprocessing efficiency of hyperspectral remote sensing image. This paper studies geometric correction parallel processing technology, aiming at windows multithread technology, OpenMP shared memory model and MPI message passing model. The principle of geometric correction parallel processing is studied, and the parallel efficiency is compared by programming on computer. The former two are applied to multi-core multi-thread processing of single machine, and the latter is used to realize multi-computer cluster system. Emphasis is placed on the advantages of OpenMP in improving data processing and CPU efficiency. In the Windows environment, based on VC programming environment, Microsoft basic class library is used to encapsulate the implementation of the algorithm. According to the principle of software design, the system structure of the software is constructed, and the functional modules of the software system, such as data pre-finishing module, radiation correction module, are designed and implemented. Geometric correction module and fast processing module etc. The geometric correction module adopts multi-thread technology, and the fast processing module provides one-click processing and batch processing functions to ensure the efficiency of data processing.
【学位授予单位】：山东大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP751

【相似文献】