Abstract:The presence of random noise in field data significantly reduces the precision of subsequent seismic processing steps. As a result, random noise suppression is essential to improve the quality of field data. Because most traditional algorithms characterize seismic data linearly, the denoising accuracy is still open to be improved. As an unsupervised deep learning method, the Deep Image Prior (DIP) algorithm can characterize seismic data nonlinearly. The DIP uses randomly generated noise as input and noisy seismic data as desired output for random noise attenuation over several rounds of training epochs. However, determining the optimal training epoch for obtaining the final denoised result of unlabeled noisy data remains a challenge. In order to terminate the DIP training in time and obtain the denoised result, we design an improved quality control criterion (IQCC) based on adjacent estimations of seismic signal. To further improve the denoising accuracy, a recursive strategy is developed that utilizes the previous desired output as the new input and the previous denoised result as the new desired output. To get the optimal denoised results using the suggested recursive algorithm, a convergence condition is also established. Numerous examples of synthetic pre-stack and post-stack data demonstrate the effectiveness of the designed IQCC and the proposed recursive strategy with a convergence condition in protecting the effective signal, especially when compared to the curvelet thresholding algorithm and the original DIP. Furthermore, the denoising accuracy is on par with that of the supervised learning algorithm, demonstrating the adaptability of the proposed recursive DIP under the convergence condition. Its superiority is further supported by field post-stack seismic data processing which uses the local similarity for performance assessments.

Regularized deep learning for unsupervised random noise attenuation in poststack seismic data

Residual Learning of Deep Convolutional Neural Network for Seismic Random Noise Attenuation

An Unsupervised Deep Learning Method for Denoising Prestack Random Noise

Attenuating Random Noise in Seismic Data by a Deep Learning Approach

Unsupervised Deep Learning for Random Noise Attenuation of Seismic Data

Self-Supervised Seismic Random Noise Attenuation With Spatial Attention From a Single Section

Unsupervised 3-D Random Noise Attenuation Using Deep Skip Autoencoder

A Fully Unsupervised and Highly Generalized Deep Learning Approach for Random Noise Suppression

Seismic Shot Gather Denoising by Using a Supervised-Deep-Learning Method with Weak Dependence on Real Noise Data: A Solution to the Lack of Real Noise Data

Unpaired training: Optimize the seismic data denoising model without paired training data

Seismic random noise suppression using deep convolutional autoencoder neural network

Unsupervised Seismic Random Noise Attenuation Based on Deep Convolutional Neural Network

Deep Learning Seismic Random Noise Attenuation via Improved Residual Convolutional Neural Network

Improving the Signal‐to‐Noise Ratio of Seismological Datasets by Unsupervised Machine Learning

Seismic random noise suppression by using deep residual U-Net

Unsupervised seismic random noise attenuation by a Recursive Deep Image Prior

Self-Supervised Seismic Swell Noise Suppression From Noisy Seismic Data

An effective self-supervised learning method for various seismic noise attenuation

A Denoising Method for Seismic Data Based on SVD and Deep Learning

Self-Supervised Attenuation Method Based on Similarity Comparisons for 3D Seismic Random Noise

Seismic Random Noise Suppression Model Based on Downsampling and Superresolution