How to Best Combine Demosaicing and Denoising?

TL;DR

This paper investigates the optimal sequence for combining demosaicing and denoising in raw image processing, concluding that demosaicing should generally precede denoising, with some exceptions at high noise levels, supported by extensive experiments.

Contribution

It provides a comprehensive analysis and practical guidelines for joint demosaicing and denoising, emphasizing low complexity algorithms over deep learning approaches.

Findings

Demosaicing first is optimal with moderate noise levels.

Partial denoising of the CFA improves results at high noise levels.

The proposed pipeline is validated on simulated and real noisy images.

Abstract

Image demosaicing and denoising play a critical role in the raw imaging pipeline. These processes have often been treated as independent, without considering their interactions. Indeed, most classic denoising methods handle noisy RGB images, not raw images. Conversely, most demosaicing methods address the demosaicing of noise free images. The real problem is to jointly denoise and demosaic noisy raw images. But the question of how to proceed is still not yet clarified. In this paper, we carry-out extensive experiments and a mathematical analysis to tackle this problem by low complexity algorithms. Indeed, both problems have been only addressed jointly by end-to-end heavy weight convolutional neural networks (CNNs), which are currently incompatible with low power portable imaging devices and remain by nature domain (or device) dependent. Our study leads us to conclude that, with moderate noise, demosaicing should be applied first, followed by denoising. This requires a simple adaptation of classic denoising algorithms to demosaiced noise, which we justify and specify. Although our main conclusion is ``demosaic first, then denoise'', we also discover that for high noise, there is a moderate PSNR gain by a more complex strategy: partial CFA denoising followed by demosaicing, and by a second denoising on the RGB image. These surprising results are obtained by a black-box optimization of the pipeline, which could be applied to any other pipeline. We validate our results on simulated and real noisy CFA images obtained from several benchmarks.