Projected Distribution Loss for Image Enhancement

TL;DR

This paper introduces a novel loss function based on aggregated 1D-Wasserstein distances between CNN features, significantly improving perceptual quality in various image enhancement tasks.

Contribution

It proposes using 1D-Wasserstein distances for feature comparison, offering a more reliable and perceptually effective loss for image enhancement models.

Findings

Outperforms existing perceptual losses in denoising, super-resolution, and other tasks.

Produces more realistic textures and details in enhanced images.

Applicable across multiple imaging frameworks.

Abstract

Features obtained from object recognition CNNs have been widely used for measuring perceptual similarities between images. Such differentiable metrics can be used as perceptual learning losses to train image enhancement models. However, the choice of the distance function between input and target features may have a consequential impact on the performance of the trained model. While using the norm of the difference between extracted features leads to limited hallucination of details, measuring the distance between distributions of features may generate more textures; yet also more unrealistic details and artifacts. In this paper, we demonstrate that aggregating 1D-Wasserstein distances between CNN activations is more reliable than the existing approaches, and it can significantly improve the perceptual performance of enhancement models. More explicitly, we show that in imaging applications such as denoising, super-resolution, demosaicing, deblurring and JPEG artifact removal, the proposed learning loss outperforms the current state-of-the-art on reference-based perceptual losses. This means that the proposed learning loss can be plugged into different imaging frameworks and produce perceptually realistic results.