To make a geologic body more outstanding, it is necessary to preserve edges when filtering. For this reason, a filtering method based on anisotropic diffusion is introduced. The key point of this method is that diffusion filtering will be made along the edge while diffusion is forbidden in the direction perpendicular to the edge, so the filtering has directionality. When doing anisotropic diffusion filtering, parameters like diffusion coefficient and threshold have great impacts on the results, so the anisotropic diffusion model parameters are discussed, the diffusion coefficient equation is introduced, and the diffusion threshold selection criterion is derived and analyzed. In addition, this method was used with the diffusion coefficient equation＇s proper diffusion threshold to highlight the tidal channel geobodies in the XX area, sand body in the YY area, and faults in the ZZ area. The good delineation results prove that the diffusion threshold selection criterion introduced in this paper is reasonable.
随机噪声的存在往往会影响地震图像分析的准确度。为了提高图像分析质量,提出了一种基于非局部均值滤波(Non Local Means)抑制地震图像随机噪声的新方法。在对滤波像素点去噪时,该方法分配给每个相似像素点的权重不依赖于2个像素点的空间距离,而是依赖以该像素点为中心的图像子块与以当前像素点为中心的子块之间的相似性,且滤波参数h的选取对滤波效果起到至关重要的作用。结合实例,对地震资料进行了具体分析。结果表明,与传统方法(如中值滤波、高斯滤波)相比,采用非局部均值滤波方法合成地震记录和实际数据时,既能有效地抑制地震随机噪声,又能较好地保留地震同相轴陡变处或同相轴弯曲处的边缘细节信息,具有实用性和有效性。
