Enhanced SVD Denoising Method Based on Hankel Matrix Decomposition for Seismic Data
Honglin Dai, Yucheng Tang, Xiangyu Han, Weiming Xiong · 2024
This study introduces an improved SVD denoising method based on the decomposition of the Hankel matrix. This method denoises by analyzing the singular value spectrum of the Hankel matrix, where larger singular values typically represent components with higher coherence in the data, and smaller singular values correspond to less coherent parts. By selecting different rank reduction parameters based on the characteristics of the processed data, this method achieves noise reduction and enhances SNR. Subsequently, a two-dimensional seismic data model is constructed to validate this algorithm. The key to this algorithm is determining the truncation position of the singular value matrix (rank reduction parameter), and the optimal rank reduction parameter is efficiently found using the energy ratio. Practical case studies have shown a significant improvement in the SNR of incoherent data, effectively suppressing random noise and improving the denoising effect for non-horizontal coherent axes. Overall, the Hankel matrix construction method effectively suppresses random noise while maintaining high fidelity of the signal data.