Random Noise Suppression of Prestack Seismic Data Using Non-Local Means via Patch Ordering in the Dual-Domain

Efficient noise removal in seismic data is crucial for accurately analysing subsurface structures because noise generated during field acquisition can considerably degrade data quality. Traditional single-domain denoising methods often struggle to preserve weak signals in prestack seismic data, potentially leading to the loss of critical information. To address this issue, we propose a novel dual-domain (DD) denoising approach called non-local means via patch ordering in DD (DD–PONLM). This method leverages the strengths of both time–space and transform domains to minimize the leakage of weak events. By employing non-local self-similarity and iterative processing in the time–space domain and discrete cosine transform domain, the proposed method effectively reduces noise while preserving weak signals. We validate the effectiveness of our method through extensive testing on both asynthetic and a field example. The results are compared with several traditional single-domain methods, demonstrating that DD–PONLM considerably improves the preservation of weak signals and reduces artefacts, such as the Gibbs phenomenon, associated with transform domain processing. This DD strategy not only enhances the signal-to-noise ratio but also preserves structural fidelity, making it a robust solution for seismic data denoising.