一种边界保持随机噪声衰减的各向异性拉普拉斯滤波技术

摘要

Since the complexity of seismic prospecting target, has brought down S/N ratio of acquired seismic data. Various residual noises in poststack seismic data can make ambiguous information of fault and lithologic boundary, thus cannot meet requirements of seismic data fine interpretation. In recent years, many researchers have paid close attention to improve fault and lithologic boundary de⁃piction, together enhanced seismic data S/N ratio edge retaining filtering technology. In allusion to uncertainty in commonly used diffu⁃sion filtering, from anisotropic diffusion filtering equation can deduce Laplace filter model. Using gradient structure tensor technology estimates seismic phase axis direction and fault position, provides anisotropic Laplace filter with specific spatial filtering structure. Us⁃ing the technology in 3D seismic data processing, can not only improve fault imaging definition, but also thin layer and very small sedi⁃mentary events imaging definition, in favor of further reservoir fine assessment.%由于地震勘探目标的复杂性，导致地震采集资料的信噪比降低，在其叠后地震资料中的各种残留噪声将会造成断层和岩性边界的信息模糊，不能满足地震资料精细解释的需求。近年来能够增强断层和岩性边界刻画，同时提高地震资料信噪比的边界保持滤波技术得到很多学者的关注。针对常用的扩散滤波方法在滤波处理时存在的不确定性，从各向异性扩散滤波方程出发，推导出拉普拉斯滤波器模型；利用梯度结构张量技术估计地震同向轴方向和断层位置，给出具有明确空间滤波结构的各向异性拉普拉斯滤波器。将该技术用于三维地震资料处理，不仅可以增强断层成像的清晰度，同时也能提高薄层和微细沉积事件的成像精度，有利于进一步做好储层精细评价。