硕士论文FIN1feng

西北工业大学硕士学位论文摘要I摘要遥感图像分割是遥感图像分析和解译中的关键技术之一，对有效利用遥感数据，进一步进行信息提取与目标识别等工作具有重要意义。高分辨率遥感图像包含丰富的颜色、纹理等信息，图像本身也含有大量噪声，因此如何较好地移除噪声，并合理利用图像所包含的各类信息对图像进行准确分割是高分辨率遥感图像分割研究中的一个经典问题。JSEG算法是一种可依次实现图像滤波、颜色量化和空间分割的彩色图像分割方法，但该方法直接用于遥感图像分割时，往往由于遥感图像中区域边界较模糊而导致对区域边界分割不准确，或由于区域内不同阴影而出现过分割现象。为了有效实现对区域边界的准确分割，本文利用能更好描述区域内颜色的同质性的局部同质矩阵校正传统JSEG算法中的局部J值，以实现对区域边界的准确反映，提高区域边界分割的准确性。为了减弱或消除传统JSEG算法的过分割现象，本文利用能稳定描述图像纹理特征的/LBPC算子进行具有相似纹理信息的颜色类图的合并。最后，采用分析法、优度实验法中的评价测度，在JSEG算法中J值和J均值计算的基础上给出了J值和J均值评价方法，对本文方法的分割效果进行评价。仿真实验结论表明本文提出的改进JSEG算法可有效克服传统的JSEG算法在高分辨遥感图像分割时存在的边界分割不准确及过分割现象。关键词:高分辨率遥感图像，JSEG算法，局部同质，/LBPC算子西北工业大学硕士学位论文AbstractIIAbstract（重新翻译）Inordertoeffectivelyanalyzeandinterprettheremotesensingdata,theremotesensingimageprocessingtechnologyhasbecomehotspot.Theremotesensingimagesegmentationisoneofkeyfactors,whichdecidethesuccessofremotesensingimageanalyzeandcalculation.Becauseonlyifthebettersegmentationeffectisobtained,thefollow-upworkscanbegetbettereffects,suchasinformationextractionandtargetrecognitionetal.Thebackgroundofthispaperishighresolutionremotesensingimage.Astheseremotesensingimagescontainawealthofcolor,textureandalotofnoiseetal,thekeypointoftheresearchishowtoeffectivelyremovenoise,smooththeimageandreasonablyusevariousinformationtosegmenttheimage.Firstly,theimagefiltering,colorspacequantizationandspatialsegmentationcanbecompletelysimultaneouslybyusetheJSEGalgorithm,whichdirectlyutilizecolorinformation.Then,InordertosolvethelimitationcaseoftheJSEG,suchasoversegmentation,theimprovedJSEGalgorithmisproposedinthispaper.Forobtainingthebettereffectofsegmentation,thelocalJvalueiscorrectedbylocalhomogeneousmatrix,whichgivesreasonablydescriptionforcolorhomogeneityinregionandtheboundaryofdifferentregion.Inaddition,forweakeningoreliminatingtheoversegmentation,theclassmapismergedbyLBP/Coperator.Itgivesstabledescriptionforthetextureofimage.Thefinally,theevaluationofimagesegmentationisanalyzedandsummarized.Inaddition,theevaluationmethodofJvalueandJaveragevalueisproposedbasedoncalculatingtheJvalueandJaveragevalueinJSEGalgorithm.Themeritisreflectedbyobjectiveevaluationdata.TheexperimentresultsshowthatthehighresolutioncolorremotesensingimageiseffectivelysegmentedbytheJSEGalgorithmandimprovedJSEGalgorithm.Inaddition,theshortagesofJSEGiseffectivelyovercomebytheimprovedmethod,forexample,theoversegmentationandinaccuratelocationofregionboundary.Keywords:Highresolutioncolorremotesensingimage,JSEGalgorithm,Localhomogeneity,LBP/Coperator西北工业大学硕士学位论文目录III目录摘要.................................................................................................................................IAbstract.............................................................................................................................II目录..............................................................................................................................III第一章绪论EquationChapter1Section1.....................................................................11.1课题研究背景及意义.....................................................................................11.1.1遥感图像成像与处理..............................................................................11.1.2遥感图像处理研究意义..........................................................................21.2遥感图像分割发展现状.................................................................................31.2.1图像分割技术发展现状..........................................................................31.2.2高分辨率遥感图像分割发展现状..........................................................41.3论文主要工作及内容.....................................................................................7第二章基于JSEG算法的彩色遥感图像分割EquationChapter2Section1..............92.1彩色特征空间...............................................................................................102.2JSEG算法......................................................................................................112.2.1彩色图像的颜色量化.............................................................................112.2.2空间分割算法........................................................................................192.3JSEG算法分割彩色遥感图像实验.............................................................262.3.1分割实验................................................................................................262.3.2实验结果及分析....................................................................................272.4本章小结.......................................................................................................28第三章基于改进JSEG算法的彩色遥感图像分割EquationChapter3Section1....293.1基于局部同质矩阵的JSEG算法................................................................293.1.1局部同质性评价....................................................................................293.1.2基于局部同质矩阵的JSEG算法改进.................................................323.1.3实验结果及分析....................................................................................323.2基于局部二进制模式的纹理描述方法.......................................................363.2.1LBP算法...........................................