Frequency Domain Loss Function for Deep Exposure Correction of Dark Images

TL;DR

This paper introduces a frequency domain loss function based on DCT/FFT for deep learning models to improve exposure correction in dark, noisy, and blurry images captured in low-light conditions, outperforming existing methods.

Contribution

The authors propose a novel multi-scale frequency domain loss function that enhances deep network training for low-light image correction, applicable to RAW and JPEG images.

Findings

Significant quantitative improvements over state-of-the-art methods.

Enhanced visual quality in corrected images.

Loss function is end-to-end differentiable and scale-agnostic.

Abstract

We address the problem of exposure correction of dark, blurry and noisy images captured in low-light conditions in the wild. Classical image-denoising filters work well in the frequency space but are constrained by several factors such as the correct choice of thresholds, frequency estimates etc. On the other hand, traditional deep networks are trained end-to-end in the RGB space by formulating this task as an image-translation problem. However, that is done without any explicit constraints on the inherent noise of the dark images and thus produce noisy and blurry outputs. To this end we propose a DCT/FFT based multi-scale loss function, which when combined with traditional losses, trains a network to translate the important features for visually pleasing output. Our loss function is end-to-end differentiable, scale-agnostic, and generic; i.e., it can be applied to both RAW and JPEG images in most existing frameworks without additional overhead. Using this loss function, we report significant improvements over the state-of-the-art using quantitative metrics and subjective tests.