随机噪声的局部正交压制方法

针对地震数据处理过程中去除的随机噪声中会残留部分有效信号的问题，提出了正则化局部正交方法，实现残余信号从随机噪声中的二次分离。首先采用传统处理方法，即具有更高压噪性能的自适应奇异值分解（ASVD）方法，将大部分信号与噪声分离；以分离出的信号为基础，利用信号与噪声的局部正交性构建目标函数，通过求解目标函数得到最优权重因子，进而计算噪声中残留的有效信号；最后将此残留有效信号与分离信号相加，得到噪声压制后的完整地震信号。模型及实际资料处理结果均表明：该方法能有效地提取分离噪声中残留的地震信号，实现了提高地震资料信噪比目的。

A local orthogonalization for seismic random noise suppression

The noise section remains a certain amount of useful signal after random noise attenuation.Therefore,we propose in this paper a new approach to retrieve useful signal from the initial noise section based on local signal-and-noise orthogonalization.We firstly use a conventional method to finish initial signal noise separation.The adaptive singular value decomposition(ASVD) is adopted for obtaining higher signal-to-noise ratio(SNR) on the initially de-noised section.Then we predict and retrieve useful signals from the initial noise via a weighting operator.The weighting operator is optimized by solving a least-squares minimization problem via shaping regularization with a smoothness constraint.The final de-noised section is composed of the retrieved signal and the initially de-noised section.Synthetic and field data examples show that the proposed approach can further retrieve useful signal from the initial noise section and effectively attenuate noise and improve seismic data SNR compared with the f-x filtering and the singular value decomposition.