地震图像序列应用一致增强性扩散方法的研究
|
摘要
针对具有线形纹理特征的地震图像序列 ,在偏微分扩散方程的基础上 ,重点研究了二维和三维的一致增强性扩散算法 .理论分析表明 ,三维一致增强性扩散算法可以利用三维地震数据体的非冗余信息 ,所以比二维的算法具有更好的性能 ,不但能更有效地提高信噪比 ,同时保护和增强边缘纹理 ,而且可以更加准确地判断实际的地质结构 ,消除原图像中的某些假象 .大量真实地震图像序列处理结果及其数据分析表明 ,无论是哪一地区的地震图像 ,在信号能量保持 80 %～ 90 %的条件下 ,三维算法提高信噪比的程度均比二维算法高 ,并可以消除原图像中局部的假断点和假连续纹线 ,使恢复的地质结构更加准确 ,有利于地震解释 ,对探明地质储量具有十分重要的意义 .
Considering that the seismic image sequences have line-like texture features, 2D and 3D coherence enhancing diffusion algorithms are emphatically researched based on partial differential diffusion equations. The theory analysis shows that 3D algorithm have better performances than 2D algorithm, because it could make use of non-redundancy information of muti-lines section images of 3D seismic data volume. So 3D algorithm could not only efficiently increase signal to noise ratio (SNR) of seismic image, as well as, protect and enhance edges and textures, but also accurately judge geological structure to eliminate some artifacts caused by reason of noise and disturbance. A great deal of experimental results and data analysis with real seismic image sequences indicated that, for seismic images obtained from whichever geological zones, 3D coherence enhancing diffusion algorithms can make the increase of SNR more than 2D algorithm. At the same time, the 80～90% of signal energy is kept. Especially, it could eliminate some local false break points and false continual line-like textures, as well as the restored geological structures are more correct. All of these are available to seismic interpretation, and they also have the very great significance for exploring the geological reserves.
Considering that the seismic image sequences have line-like texture features, 2D and 3D coherence enhancing diffusion algorithms are emphatically researched based on partial differential diffusion equations. The theory analysis shows that 3D algorithm have better performances than 2D algorithm, because it could make use of non-redundancy information of muti-lines section images of 3D seismic data volume. So 3D algorithm could not only efficiently increase signal to noise ratio (SNR) of seismic image, as well as, protect and enhance edges and textures, but also accurately judge geological structure to eliminate some artifacts caused by reason of noise and disturbance. A great deal of experimental results and data analysis with real seismic image sequences indicated that, for seismic images obtained from whichever geological zones, 3D coherence enhancing diffusion algorithms can make the increase of SNR more than 2D algorithm. At the same time, the 80～90% of signal energy is kept. Especially, it could eliminate some local false break points and false continual line-like textures, as well as the restored geological structures are more correct. All of these are available to seismic interpretation, and they also have the very great significance for exploring the geological reserves.
引文
1IijimaT ..Basictheoryofpatternnormalization(forthecaseofatypicalone dimensionalpattern).BulletinoftheElectrotechnicalLaboratory,1962,26:368～388(inJapanese)
2WitkinA .P ..Scale spacefiltering.In:ProceedingsoftheEighthInternationalJointConferenceonArtificialIntelligence,Karl sruhe,WestGermany,PaloAlto,California,1983,1019～1022
3KoenderinkJ..Thestructureofimages.BiologicalCybernation,1984,50:363～370
4PeronaP .,MalikJ..Scale spaceandedgedetectionusinganisotropicdiffusion.In:ProceedingsoftheIEEEComputerSoci etyWorkshoponComputerVision,IEEEComputerSocietyPress,Miami,1987,16～27
5PeronaP .,MalikJ..Scale spaceandedgedetectionusinganisotropicdiffusion.IEEETransactionsonPatternAnalysisandMachineIntelligence,1990,12(7):629～639
6GilboalG .,SochenN .etal..Forward and backwarddiffusionprocessesforadaptiveimageenhancementanddenoising.IEEETransactionsonImageProcessing,2002,11(7):689～703
7Gilboa1G .,SochenN .etal..Regularizedshockfiltersandcom plexdiffusion.In:ProceedingsoftheECCV ,Copenhagen,2002,399～413
8WeickertJ..Multiscaletextureenhancement.In:Proceedingsofthe6thInternationalConferenceonComputerAnalysisofImagesandPatterns,PragueCzech,1995,230～237
9LuJiMengetal..TheoryofSeismologyExplorationandDataInterpretation.Beijing:TheCenterofExplorationTrainingofPetroleumIndustryDepartment,1982(inChinese)(陆基孟等.地震勘探原理及资料解释.北京:石油工业部勘探培训中心,1982)
10WeickertJ..Areviewofnonlineardiffusionfiltering.In:HaarRomeryB .T .etal.eds.ScaleSpaceTheoryinComputerVi sion,LectureNotesinComputerScience1252,Berlin:Springer,1997,3～28
11WeickertJ..Coherence enhancingdiffusionfiltering.Internation alJournalofComputerVision,1999,31(2/3):111～127
12Bahn.MarkM ..Alinearrelationexitsamongbraindiffusioneigenvaluesmeasuredbydiffusiontensormagneticresonanceimag ing.JournalofMagneticResonance,1999,137(1):33～38
13WeickertJ..Aschemeforcoherence enhancingdiffusionfilteringwithoptimizedrotationinvariance.JournalofVisualCommunica tionandImageRepresentation,2002,13(1/2):103～118
14ChenFeng,LiJinZongetal..Amodelofnonlinearanisotropicdiffusionforincreasingsignaltonoiseratioofseismicimage.PetroleumExplorationandDevelopment,2004,31(2):77～80(inChinese)(陈 凤,李金宗等.一种非线性各向异性扩散提高地震图像信噪比的应用.石油勘探与开发,2004,31(2):77～80)
2WitkinA .P ..Scale spacefiltering.In:ProceedingsoftheEighthInternationalJointConferenceonArtificialIntelligence,Karl sruhe,WestGermany,PaloAlto,California,1983,1019～1022
3KoenderinkJ..Thestructureofimages.BiologicalCybernation,1984,50:363～370
4PeronaP .,MalikJ..Scale spaceandedgedetectionusinganisotropicdiffusion.In:ProceedingsoftheIEEEComputerSoci etyWorkshoponComputerVision,IEEEComputerSocietyPress,Miami,1987,16～27
5PeronaP .,MalikJ..Scale spaceandedgedetectionusinganisotropicdiffusion.IEEETransactionsonPatternAnalysisandMachineIntelligence,1990,12(7):629～639
6GilboalG .,SochenN .etal..Forward and backwarddiffusionprocessesforadaptiveimageenhancementanddenoising.IEEETransactionsonImageProcessing,2002,11(7):689～703
7Gilboa1G .,SochenN .etal..Regularizedshockfiltersandcom plexdiffusion.In:ProceedingsoftheECCV ,Copenhagen,2002,399～413
8WeickertJ..Multiscaletextureenhancement.In:Proceedingsofthe6thInternationalConferenceonComputerAnalysisofImagesandPatterns,PragueCzech,1995,230～237
9LuJiMengetal..TheoryofSeismologyExplorationandDataInterpretation.Beijing:TheCenterofExplorationTrainingofPetroleumIndustryDepartment,1982(inChinese)(陆基孟等.地震勘探原理及资料解释.北京:石油工业部勘探培训中心,1982)
10WeickertJ..Areviewofnonlineardiffusionfiltering.In:HaarRomeryB .T .etal.eds.ScaleSpaceTheoryinComputerVi sion,LectureNotesinComputerScience1252,Berlin:Springer,1997,3～28
11WeickertJ..Coherence enhancingdiffusionfiltering.Internation alJournalofComputerVision,1999,31(2/3):111～127
12Bahn.MarkM ..Alinearrelationexitsamongbraindiffusioneigenvaluesmeasuredbydiffusiontensormagneticresonanceimag ing.JournalofMagneticResonance,1999,137(1):33～38
13WeickertJ..Aschemeforcoherence enhancingdiffusionfilteringwithoptimizedrotationinvariance.JournalofVisualCommunica tionandImageRepresentation,2002,13(1/2):103～118
14ChenFeng,LiJinZongetal..Amodelofnonlinearanisotropicdiffusionforincreasingsignaltonoiseratioofseismicimage.PetroleumExplorationandDevelopment,2004,31(2):77～80(inChinese)(陈 凤,李金宗等.一种非线性各向异性扩散提高地震图像信噪比的应用.石油勘探与开发,2004,31(2):77～80)
版权所有：© 2023 中国地质图书馆 中国地质调查局地学文献中心