Speckle Image Restoration without Clean Data

TL;DR

This paper introduces a novel speckle noise removal algorithm that operates without clean data or multiple observations, effectively handling spectral images and real-world holography samples with superior results.

Contribution

The proposed method uniquely removes speckle noise without clean data or multiple observations, and works without prior noise distribution knowledge, especially effective for spectral images.

Findings

Superior quantitative and visual results compared to baselines

Effective on real-world digital holography samples

Robust across different speckle noise strengths

Abstract

Speckle noise is an inherent disturbance in coherent imaging systems such as digital holography, synthetic aperture radar, optical coherence tomography, or ultrasound systems. These systems usually produce only single observation per view angle of the same interest object, imposing the difficulty to leverage the statistic among observations. We propose a novel image restoration algorithm that can perform speckle noise removal without clean data and does not require multiple noisy observations in the same view angle. Our proposed method can also be applied to the situation without knowing the noise distribution as prior. We demonstrate our method is especially well-suited for spectral images by first validating on the synthetic dataset, and also applied on real-world digital holography samples. The results are superior in both quantitative measurement and visual inspection compared to several widely applied baselines. Our method even shows promising results across different speckle noise strengths, without the clean data needed.