Preventing Overfitting in Deep Image Prior for Hyperspectral Image Denoising

TL;DR

This paper introduces a novel method to prevent overfitting in deep image prior-based hyperspectral image denoising by combining data fidelity with sensitivity regularization, leading to improved denoising results.

Contribution

It proposes a joint approach using robust data fidelity and explicit regularization to mitigate overfitting in DIP for hyperspectral image denoising.

Findings

Effective overfitting prevention in DIP-based HSI denoising.

Superior denoising performance on real HSIs with various noise types.

Outperforms existing DIP-based methods.

Abstract

Deep image prior (DIP) is an unsupervised deep learning framework that has been successfully applied to a variety of inverse imaging problems. However, DIP-based methods are inherently prone to overfitting, which leads to performance degradation and necessitates early stopping. In this paper, we propose a method to mitigate overfitting in DIP-based hyperspectral image (HSI) denoising by jointly combining robust data fidelity and explicit sensitivity regularization. The proposed approach employs a Smooth $\ell_1$ data term together with a divergence-based regularization and input optimization during training. Experimental results on real HSIs corrupted by Gaussian, sparse, and stripe noise demonstrate that the proposed method effectively prevents overfitting and achieves superior denoising performance compared to state-of-the-art DIP-based HSI denoising methods.