Acta Petrolei Sinica ›› 2016, Vol. 37 ›› Issue (4): 490-498,507.DOI: 10.7623/syxb201604008

• Petroleum Exploration • Previous Articles     Next Articles

A combined denoising method based on Curvelet transform and anisotropic diffusion filtering

Sun Chengyu1, Sun Chengyu1, Shi Xiaolei2, Liu Mutian1   

  1. 1. School of Geoscience, China University of Petroleum, Shandong Qingdao 266580, China;
    2. Xinjiang Golden Gobi Oil Sands Development Company Limited, Xinjiang Karamay 834000, China
  • Received:2015-11-02 Revised:2016-03-01 Online:2016-04-25 Published:2016-05-06

基于曲波变换和各向异性扩散滤波的联合去噪技术

姚振岸1, 孙成禹1, 石小磊2, 刘牧天1   

  1. 1. 中国石油大学地球科学与技术学院 山东青岛 266580;
    2. 新疆金戈壁油砂矿开发有限责任公司 新疆克拉玛依 834000
  • 通讯作者: 姚振岸,男,1990年10月生,2013年获中国石油大学(华东)学士学位,现为中国石油大学(华东)博士研究生,主要从事地震波传播理论、地震资料处理和储层岩石物理等方面的研究。Email:an6428060@163.com
  • 作者简介:姚振岸,男,1990年10月生,2013年获中国石油大学(华东)学士学位,现为中国石油大学(华东)博士研究生,主要从事地震波传播理论、地震资料处理和储层岩石物理等方面的研究。Email:an6428060@163.com
  • 基金资助:

    国家自然科学基金项目(No.41374123,No.41504097)和山东省自然科学基金项目(ZR2013DQ020)资助。

Abstract:

Curvelet transform threshold denoising method is an effective way for random noise suppression, of which the denoising effect depends on threshold selection, and two inherent defects exist, i.e., excessive smoothing and surrounding effect. Anisotropic diffusion filtering method is able to preserve the border structures based on thermal diffusion partial differential equation, applicable to process the seismic texture images with high signal to noise ratio for image enhancing. Based on the analysis of Curvelet transform threshold, a new effective combined denoising algorithm is presented to fully integrate the advantages of both denoising methods. On the one hand, it avoids the difficult in Curvelet threshold selection. On the other hand, the border and amplitude are preserved to improve the defect of Curvelet transform threshold method. Through pre-stack and post-stack simulations on seismic data and actual seismic dada tests, it has been proven that such method can not only remove random noises, but also preserve the amplitude and morphological characteristics of seismic events effectively. Six quantitative evaluation indices are used to analyze the advantages of such method in improving signal to noise ratio and preserving border characteristics compared with multiple Curvelet threshold desnoising method.

Key words: Curvelet transform, anisotropic diffusion, random denoising, border and amplitude preserving, wave filtering

摘要:

曲波变换阈值去噪方法是一种有效的随机噪音压制方法,其去噪效果取决于阈值的选取,且存在过度平滑和环绕效应2种固有缺陷。各向异性扩散滤波方法是一种基于热扩散偏微分方程的边界保持的滤波方法,适用于信噪比较高的地震纹理图像,往往用于图像增强。基于对曲波变换阈值的分析,充分结合2种去噪方法的优点,提出了一种有效的联合去噪算法,避开了曲波阈值选取的困难,实现了边界和振幅的保持,改善了曲波变换阈值方法的缺陷。通过对叠前、叠后模拟地震数据以及实际地震数据测试,证明了该方法在去除随机噪音的同时能有效保持地震同相轴的振幅和形态特征。通过6种定量化指标,分析了该方法相对于多种曲波阈值去噪方法在提高信噪比和保持边界特征方面的优势。

关键词: 曲波变换, 各向异性扩散, 随机噪音去除, 边界和振幅保持, 滤波

CLC Number: