【作者】 刘冬；

【导师】 杨庆芳；

【作者基本信息】 吉林大学 ， 交通信息工程及控制， 2011， 硕士

【摘要】 随着航空航天技术、电子传感器技术以及通信技术的高速发展,遥感技术已经发展到了一个崭新的阶段。现代的遥感技术具有具有覆盖范围广、采集速度快、实时性好、精度高、可大范围重复性观测等优势。高空间的分辨率已经达到了纳米级别,观测波段数已经达到了数百个。遥感图像中蕴藏着丰富的地理数据信息。随着空间技术的飞速发展,遥感图像的获取成本不断的降低,分辨率不断提高,采集信息量急剧增加。遥感图像已经成为获取地理信息数据的重要途径。道路网是地理信息系统中一个重要的组成部分,在很多领域都有重要的应用价值。例如电子地图制作和更新、空间数据库的建立和更新、城市交通管理、城市建筑规划、汽车导航、军事目标攻击等等。鉴于遥感图像的覆盖范围广、获取速度快、更新周期短、蕴藏数据量丰富等特点,高分辨率遥感图像是获取道路网络信息的良好来源。研究如何从高分辨率遥感图像中提取路网信息将进一步推动遥感图像处理技术在人们的生产生活中的应用。数学形态学(MathematicalMorphology)诞生于1964年,由法国科学家马瑟荣(G. Matheron)和其学生塞拉(J. Serra)共同创立。数学形态学是一门建立在严格数学理论基础上的学科,其基本思想和方法对图像处理的理论和技术产生了重大的影响。数学形态学是一项以图像形态几何特征及结构特征的定量描述与分析为主要研究内容的新兴图像分析技术,是线性向非线性处理的延拓。形态变换最大的特点是能将大量复杂的影像处理运算转换成为最基本的移位与逻辑运算的复合来完成,这是其方法在图像处理的诸多领域能越来越广泛地被人们所关注、挖掘和应用,形态分析的理论、方法及其与之相关的诸多技术能不断地发展、深入和创新的重要原因。如何将这样一项新兴发展的技术有效地应用于遥感影像信息的处理中,使遥感技术能更好地发挥作用是一项有价值的研究课题。随着空间信息技术的飞速发展,高分辨率遥感图像中所蕴含的信息量急剧增加,一幅高分辨率遥感图像所占用的存储空间往往达到十几G甚至几十G,随之而来的就是处理这样的遥感图像所需要进行的巨大的运算量。并行计算技术近年来应用越来越广泛,它的目标定位是在短时间内解决大规模的或计算量巨大的科学计算问题。因此可以应用并行计算技术来解决海量图像数据处理问题。本论文的研究目标是开发一套从高分辨率的遥感图像中准确的提取出感兴趣道路信息的技术方法,具体应用数学形态学理论中的灰度形态学部分以及其他数字图像处理理论,并应用分布式并行计算技术实现这一技术方法的并行化流程。本文由两个重点研究内容：第一,深入研究分析数学形态学理论,探讨其在遥感图像处理领域的应用,并且应用此理论和其它数字图像处理理论研发高效的高分辨率遥感图像道路提取技术方法；第二,探讨研究MPI(分布式消息传递并行编程模型)和OPENMP(共享存储式并行编程模型)的应用特点,重点研究实现MPI+OPENMP混合编程模型的应用模式,并且应用MPI+OPENMP混合编程模型实现前面研发的基于数学形态学的高分辨率遥感图像道路提取的并行实现方法。更多还原

【Abstract】 With the rapid development of aerospace technology, electronic sensor technology and communications technology, remote sensing technology is reached a new stage. Modern remote sensing technology has great advantages at many aspects, for example, Broad coverage, fast and timely information collection, high precision data. High altitude resolution has reached nano-level, and there are about hundreds wave bands that can be observed.remote sensing images have huge Geographic data information. With the rapid development of space technology, acquisition costs of remote sensing images is lower, resolution is higher, collecting information rapid increase. Remote sensing image has become a important way to get GIS data. Road network is a important part of GIS, which has important application value in many researchs,for example, road network is applied to produce and update electronic map,to create and update spatial database,to manage urban transport,to plan urban construction,to navigate vehicles and to attack military objectives. Because of remote sensing image’s advantages, high resolution remote sensing image is a well source to get road network informations.how to get road network informations from high resolution remote sensing image, the research can further promote application of remote sensing image processing technology in people’s production and life.Mathematical morphology was born in 1964.Scientists from France G.Matheron and his student J.Serra co-founded it. Mathematical morphology is the establishment of a strict mathematical theory on the basis of disciplines.Its basic ideas and menthods of image processing theory and technology have had a significant impact.Mathematical morphology is considered to be a fresh technique for image analysis with strong mathematical flavor,in particular for those applications where image geometric shape and structure characteristics quantitative analysis description aspectd are relevant.It is a powerful nonlinear tool for extending by linear approach.One of the most important features of morphological transformation is that it can compele the mass complex image processing poerations by using basic shift and logic operation.That it can complete the mass complex image processing operations by using basic shift and logic operation.That is why the morphological method brings much great interest,attracts so many people to research its theory and arithmetic,and at the same time,stands as an established image analysis technique with applications in almost all science and engineering disciplines dealing with digital spatial data,which also let morphological theory and techniques ge deep into development and innovation.Accordingly,it is important and significant to study how to use this new technique to geoscience and remote sensing,and make the remote sensing technique to be more powerful. With the rapid development of space technology, high resolution remote sensing image information rapid increase.The storage space of a high resolution remote sensing image is more than 10 Gb, so the computation is huge when processing this remote sensing Image. In recent years, application of parallel computing is more and more extensive, which objective is to solve large scale and huge computation scientific computing problems in short time.So, parallel computing can be used to process image of massive data.The paper’s objective is to R&D a technical method that can be used to get useful road informations from high resolution remote sensing images. Mathematical morphology theory and other digital image processing theorys are uesd in this technical method,and then,use parallel computing technology to parallel execution above technical method.there are two research focus:first, in-depth research and analysis mathematical morphology theory, discussion the theory’s application in remote sensing image processing, R&D efficient method of getting useful road informations from high resolution remote sensing images;Sedcond, In-depth analysis application features of MPI(Message Passing Interface) and OPENMP(Shared Memory Parallel Programming Model), focus on achieving the application mode of MPI+OPENMP hybrid programming model, further, appling MPI+OPENMP hybrid programming model to parallel execution above technical method that is proposed to get useful road informations from high resolution remote sensing images.更多还原