基于DeCNN的逆时偏移低频噪声压制方法

Eliminating Low-Frequency Noise in Reverse-Time Migration Based on DeCNN

逆时偏移由于其高分辨率的成像效果而得到广泛应用。但双程波动方程波场延拓在互相关成像条件下会产生较强的低频噪声,影响成像质量。本文构建基于U-Net的卷积-反卷积神经网络(DeCNN)压制逆时偏移中的低频噪声。以含低频噪声的震源归一化成像结果作为训练数据,以Laplace滤波结果作为标签,基于数据驱动获得神经网络模型。模型试算和迁移学习对比结果表明,DeCNN有较好的去噪效果,相比于U-Net,其在迁移的SEG/EAGE标准盐丘模型和Marmousi模型上均有更好的噪声压制效果。与常规上下行波分离的低频噪声压制方法相比,DeCNN在训练完成后,将震源归一化成像结果输入网络中,可以在数秒内输出高分辨率成像结果,效率远高于常规波场分离低频噪声压制方法,并且在噪声压制效果方面具有一定优势。

Reverse-time migration(RTM)is widely employed for its ability to produce high-resolution imaging results.Nevertheless,wavefield extrapolation based on the two-way wave equation often leads to pronounced low-frequency noise under cross-correlation imaging condition.This issue significantly impacts the quality of the resulting images.In this study,we introduce a convolution-deconvolution neural network(DeCNN)built upon the U-Net architecture to mitigate the presence of low-frequency noise in RTM.We utilize source-normalized imaging results with low-frequency noise as training data,and the Laplace filtering results as labels to abtain the neural network model,grounded in a data-driven approach.The comparison results of model trials and transfer learning demonstrations highlight the superior denoising proficiency of DeCNN.It outperforms U-Net,effectively suppressing noise in scenarios such as the transfer SEG/EAGE standard salt dome model and Marmousi model.In contrast to conventional techniques aimed at mitigating low-frequency noise through up and down traveling wave decomposition,DeCNN can input the source-normalized imaging results into the network after training,and the network can output high resolution imaging results in seconds.The efficiency is much higher than that of conventional low-frequency noise suppression methods using wavefield decomposition,and it has certain advantages in noise suppression effect.