基于改进K-SVD字典学习方法的地震数据去噪

Seismic data denoising based on improved K-SVD dictionary learning method

为实现更好的地震数据去噪技术,笔者引入一种新的算法:快速迭代收缩阀值法(FISTA),通过FISTA和K-奇异值分解(K-SVD)不断迭代更新K-SVD字典,利用更新得到的K-SVD字典对地震数据进行稀疏表示,去除稀疏系数中较小的数值,使数据中的随机噪声得到压制。对层状模型合成地震记录,Marmousi模型合成地震记录以及实际地震数据进行对比实验,得出FISTA算法较OMP算法能更好地提高地震数据的信噪比,同时有效地保护了反射信号。

In order to achieve better seismic data denoising technology,a new algorithm is introduced in this paper：fast iterative shrinkage-thresholding algorithm（FISTA）.The K-SVD dictionary is iteratively updated by FISTA and K-singular value decomposition（K-SVD）.The updated K-SVD dictionary sparsely represents the seismic data and removes the smaller sparse coefficients,which suppresses the random noise in the data.Comparison among simulation data,Marmousi model seismic data and actual seismic data indicates that FISTA algorithm can improve signal-to-noise ratio of seismic data and protect reflection signal more effectively than orthogonal matching pursuit（OMP） algorithm.