Non-local means (NLM) is one of the classical patch-based methods for random noise attenuation. It assumes that there is a lot of redundant information existing in similar patches, which can be utilized to restore the original data. However, this method is computationally expensive due to a large number of overlapping patches. Besides, since this method uses a weighted average of patches to suppress noise, when applied to the data with complicated structure, the “average effect” may appear in the denoised results. We propose to implement the NLM in frequency-space domain, which can be called the adaptive frequency-domain non-local means (AFNLM). This novel strategy will significantly reduce the computational cost and improve the quality of the final result compared to the traditional NLM. Considering the impact of filtering parameters on the final result, we also build a mapping relationship between the noise strength and filtering parameters, which could help obtain the denoised result with less signal leakage. We provide both synthetic and field examples to show the superiority of this method over the traditional f– x prediction and NLM methods.
Seismic Random Noise Attenuation via Non-Local Means Filters
