提高地震图像信噪比的非线性各向异性扩散算法模型

英文篇名：A model of nonlinear anisotropic diffusion for increasing signal to noise ratio of seismic image
中文刊名：石油勘探与开发
英文刊名：Petroleum Exploration and Development
作者：陈凤 ; 李金宗 ; 李冬冬 ; 陈惠平
英文作者：CHEN Feng~ (1) ; LI Jin-zong~ (1) ; LI Dong-dong~ (1) ; CHEN Hui-ping~ (2) (1. Harbin Institute of Technology ; Heilongjiang 150001 ; China ; 2. Daqing Oilfield Co. Ltd. ; PetroChina ; Heilongjiang 163001 ; China)
中文关键词：地震图像 ; 线形纹理特征 ; 偏微分方程 ; 非线性各向异性扩散 ; 一致增强性扩散 ; 信噪比
英文关键词：seismic image ; line-like texture characters ; partial differential equation ; nonlinear anisotropic diffusion ; coherence enhancing diffusion ; signal to noise ratio
出版日期：2004-04-23
机构：哈尔滨工业大学 ,哈尔滨工业大学 ,哈尔滨工业大学 ,中国大庆油田有限公司
年：2004
期：02
出版单位：石油勘探与开发

摘要

原始地震图像属于线形纹理图像 ,地震同相轴表现为线条结构。选取基于偏微分方程的非线性各向异性扩散 (即一致增强性扩散 )算法模型 ,通过结构张量的特征向量控制扩散方向 ,利用扩散张量的特征值控制特征方向上的扩散量 ,对原始地震图像进行图像处理 ,增强具有一致性的同相轴及其所反映的重要地质结构。利用能谱分析设计了地震剖面图像信噪比的计算公式 ,确定了扩散参数的选择范围。处理实际地震剖面图像的结果表明 ,在信号结构关系不变及其能量保持 80 %～ 90 %的条件下 ,原高信噪比地震图像的信噪比可增加约 9dB ,原低信噪比地震图像的信噪比增加大约 15dB ,并且不会出现假像 ,增强了同相轴及重要的地质结构。图 1表 1参 10
According to the line-like texture characters of seismic images, an algorithm of coherence enhancing diffusion, which is a kind of nonlinear anisotropic diffusion based on partial differential equation, is proposed.The original seismic images are disposed by controlling diffusion direction by the eigenvector of the structure tensor and controlling the diffusion amount in the characteristic direction by the eigenvalue of diffusionmeasure ment. Energy spectrum analysis is used to develop the formula of the signal to noise ratio (SNR) and the ranges of the diffusion parameters are also obtained. Processing results of real seismic images show that, under the condition of constant structure relation and signal energy of 80%-90%, the SNR increases of seismic images with high SNR and low SNR are about 9dB and 15dB, respectively. Also, there won't be false images, and seismic events and important geological information are enhanced.

引文

[1]　谢里夫RE ,等.勘探地震学[M].北京:石油工业出版社,1999.134.
    [2]　JoachimWeickert.AReviewofnonlineardiffusionfiltering[A].LectureNotesinComputerScience[C].Springer,Berlin,1997.328.
    [3]　PeronaP ,MalikJ.Scale spaceandedgedetectionusinganiso tropicdiffusion[A].IEEEComp.Soc.WorkshoponComputerVision[C].MiamiBeach,1987.1622.
    [4]WitkinAP .Scale spacefiltering[A].InternationalJointConferenceonArtificialIntellegence[C].Kalsruhe,WestGermany,1983.10191022.
    [5]PeronaP ,MalikJ.scale spaceandedgedetectionusinganisotropicdiffusion[J].IEEETransactionsonPatternAnalysisandMachineIntelligence,1990,12(7):629639.
    [6]　JoachimWeickert.Multiscaletextureenhancement[A].Proc6thInternationalConferenceonComputerAnalysisofImagesandPatterns[C].PragueCzech,1995.230237.
    [7]　JoachimWeickert.Coherence enhancingdiffusionfiltering[J].Inte rnationalJournalofComputerVision,1999,31(2/3):111127.
    [8]JoachimWeickert,HannoScharr.Aschemeforcoherence enhancingdiffusionfilteringwithoptimizedrotationinvariance[J].JournalofVisualCommunicationandImageRepresentation,2002,12(1/2):103118.
    [9]　DeguchiK ,etal.Detectionandenhancementoflinestructuresinanimagebyanisotropicdiffusion[J].PatternRecognitionLetters,2002,23(12):13991405.
    [10]　陆基孟,等.地震勘探原理及资料解释[Z].北京:石油工业部勘探培训中心,1982.