# Can learning from natural image denoising be used for seismic data interpolation?

## TL;DR

This paper introduces a CNN denoising-based seismic data interpolation method that leverages natural image denoising training, reducing the need for seismic training data and demonstrating high generalizability and effective performance on synthetic and field data.

## Contribution

The paper presents a novel seismic interpolation approach using CNN denoisers trained on natural images, integrated into POCS, which requires no seismic-specific training data.

## Key findings

- High interpolation quality on synthetic data
- Effective de-aliasing and weak-feature reconstruction on field data
- Outperforms traditional filtering and transform methods

## Abstract

We propose a convolutional neural network (CNN) denoising based method for seismic data interpolation. It provides a simple and efficient way to break though the lack problem of geophysical training labels that are often required by deep learning methods. The new method consists of two steps: (1) Train a set of CNN denoisers from natural image clean-noisy pairs to learn denoising; (2) Integrate the trained CNN denoisers into project onto convex set (POCS) framework to perform seismic data interpolation. The method alleviates the demanding of seismic big data with similar features as applications of end-to-end deep learning on seismic data interpolation. Additionally, the proposed method is flexible for many cases of traces missing because missing cases are not involved in the training step, and thus it is of plug-and-play nature. These indicate the high generalizability of our approach and the reduction of the need of the problem-specific training. Primary results on synthetic and field data show promising interpolation performances of the presented CNN-POCS method in terms of signal-to-noise ratio, de-aliasing and weak-feature reconstruction, in comparison with traditional $f$-$x$ prediction filtering and curvelet transform based POCS methods.

---
Source: https://tomesphere.com/paper/1902.10379