【摘要】 油气采集过程给地震数据带来大量高斯随机噪声，非局部均值降噪算法降噪后同相轴边缘过度平滑。为此提出一种结合边缘检测和非局部均值的地震数据降噪算法（Sobel8-NLM），通过八方向Sobel算子准确提取同相轴边缘，并改进非局部均值算法的权值函数，使地震数据中结构相差小的邻域权值不变，结构相差大的邻域权值变小，在降噪的同时有效提升信噪比及结构相似度。对合成地震数据及实际地震数据分别降噪，并与非局部均值算法（NLM）、结合Sobel算子的非局部均值算法（Sobel2-NLM）、结合Canny算子的非局部均值算法（Canny-NLM）进行对比，采用峰值信噪比、均方误差、平均结构相似度指标，验证了算法的有效性和可行性。更多还原

【Abstract】 The process of oil and gas acquisition brings a lot of random noise to seismic data. After denoising by non-local mean denoising algorithm, the edge of coaxial axis is excessively smooth. In this paper, a new seismic data denoising algorithm(Sobel 8-NLM) is proposed, which combines the edge detection and non-local mean. The eight-direction Sobel operator is used to accurately extract the edge of the coaxial line, and the weight function of the non-local mean algorithm is improved to keep the neighborhood weights of the small structural difference in seismic data unchanged, and the neighborhood weights of the large structural difference become smaller, so that the signal-to-noise ratio and the structural similarity can be improved while reducing noise. The synthetic seismic data and actual seismic data are de-noised separately,and compared with non-local mean algorithm(NLM), non-local mean algorithm(Sobel2-NLM) combined with Sobel operator and non-local mean algorithm(Canny-NLM) combined with Canny operator. The effectiveness and feasibility of the algorithm are verified by peak signal-to-noise ratio(PSNR), mean square error(MSE) and average structural similarity index.更多还原