ISHANDONGUNIVERSITYOFTECHNOLOGY毕业论文矩阵理论在数字图像处理中的应用学院:理学院专业:数学与应用数学(师范类)学生姓名:刘小慧学号:1112124027指导教师:曹永林2015年6月中文摘要I摘要矩阵作为研究数学问题的一项基础工具,有着自身特有的性质和运算方法,它不仅可以对不同的问题进行针对性简化,还可以快速看到问题的本质并加以解决。计算机对图像进行处理和显示的基础是数字图像,而数字图像的本质是nm(每行m个像素,总共n行)的矩阵。从而,便可以通过像素矩阵把图像处理归结到矩阵分析的方法中来,利用分析矩阵的方式来对图像进行相应的处理,实现图像处理与矩阵分析的融合。首先,本文介绍了数字图像处理的目的、意义以及在社会生活和科学研究中各方面的应用,其主要涉及航天和航空技术、生物医学、军事公安等方面。在第二章重点介绍了由连续图像获取数字图像的方法,该方法主要包括采样和量化两个过程。在数字图像的基础上,本文主要实现了以下几个处理:(1)利用图像的滤波理论,实现图像去噪,改善图像的质量;(2)利用矩阵的初等变换理论,实现了图像的几何变换,主要包括平移变换、旋转变换和镜像变换;(3)先从集合角度介绍了形态学的基本运算,又结合其几何意义加以深化理解。此外,本文重点探讨了矩阵的非负分解理论,分解矩阵的目的是从图像中提取有效信息。通过对几种矩阵分解方法的比较,最终发现,基于最小二乘法的非负矩阵分解法的分解结果更具有实用性。最后,本文将非负矩阵分解理论应用到人脸识别技术处理中,通过与主成分分析法比较发现,非负矩阵分解法因有了非负控制,其对人脸特征的提取更具有直观意义上的部分合成整体的效果,物理意义也更加明显。矩阵的出现不但简化了方程求解的过程,而且对现实生活也有理论指导意义。通过矩阵理论,我们可以满足计算机处理图像的要求,实现对数字图像的变换和处理,使人脸识别技术原理更直观。同时,通过这些理论让我们更清楚的知道,科学理论是科学实践的基础,数学作为一门基础学科,为其他应用科学提供了坚实的理论基础。关键字:矩阵理论,数字图像,非负矩阵分解,人脸识别英文摘要IIAbstractAsabasictooltostudymathematicsproblem,matrixhasitsownuniquepropertiesandoperationmethod.Notonlycanitsimplifythedifferentproblems,butalsocanseethenatureoftheproblemtosolveitquickly.ThebasicofComputer’sprocessinganddisplayingtotheimageisdigitalimage,andtheessenceofdigitalimageisamatrixofm*n(Mpixelsperline,atotalofNmatrix).Thus,wecouldattributedtheprocessingoftheimagetomatrixanalysisthroughpixelmatrix,andprocessingimagebythewayofanalysismatrix,approachtothefusionbeteenimageandmatrix.Firstly,thispaperintroducestheobjectiveandsignificanceofthedigitalimageprocessing,andtheapplicationinallaspectsofsociallifeandscientificresearchalso,whichmainlyrelatestotheaerospaceandaviationtechnology,bio-medicine,militaryandotheraspectsofpublicsecurity.Inthesecondchapteritmainlyintroducesthemethodofacquiringdigitalimagesfromcontinuousimage,whichincludetwoprocesses:samplingandquantification.Thispaper,onthebaseofdigitalimageprocessing,mainlycarryoutthefollowing:(1)usingimagefiltertheory,realizedimageremove-noising,improvedtheimagequality;(2)usingthetheoryofelementarytransformationofmatrix,realizedimagegeometrictransformation,includingtranslation,rotation,mirroring;(3)introducesthebasicmorphologicaloperationsinthesightoftheset,then,combinedwiththegeometricsignificancetoresolvedepth.Then,thispaperfocusesonthenon-negativematrixfactorizationtheory,whichtobringeffectiveinformationfromtheimage.Throughthecomparisonofseveralmatrixdecompositionmethods,findoutintheendthattheresultofthenon-negativematrixfactorizationthatbasedontheleastsquaremethodismorepractical.Finally,thenon-negativematrixfactorizationtheoryisappliedintothefacerecognitiontechnology,andthroughthecomparisonwithprincipalcomponentanalysismethod,wecouldfindoutthatnon-negativematrixfactorizationmethodismoreintuitivefortheextractionofhumanfacialfeaturesandmoreobviousinphysicalmeaning.Thematrixnotonlysimplifiestheprocessofsolvingequations,butalsohastheoreticalsignificanceonthereallife.Throughmatrixtheory,wecanmeettherequirementofimageprocessingincomputer,andrealizethetransformationandprocessingtodigitalimage,maketheprincipleoffacerecognitiontechnologymoreintuitive.Atthesametime,throughthesetheories,weknowmoreclearlythatthescientifictheoryisthefoundationofscientificpractice.Asabasicsubject,mathematicscanprovideasolidtheoreticalbasisforotherappliedsciences.Keywords:matrixtheory,digitalimage,non-negativematrixfactorization,facerecognition目录III目录摘要..............................................................I摘要...............................................................................................................................................IAbstract...........................................................................................................................................II目录............................................................................................................................................III第一章概述..........................................................................................................................-1-1.1数字图像的基本概念...............................................................................................-1-1.1.1图像及数字图像...........................................................................................-1-1.1.2数字图像处理................................................................................................-1-1.1.3数字图像处理的目的...................................................................................-2-1.1.4数字图像处理的主要内容...........................................................................-2-1.2数字图像处理的应用及发展..................................................................................-3-1.2.1数字图像处理技术的应用...........................................................................-3-1.2.2数字图像处理技术的发展...........................................................................-4-第二章图像数字化..................................................................................................................-5-2.1连续图像的数学描述...............................................................................................-5-2.2图像