Underwater image enhancement with Image Colorfulness Measure

TL;DR

This paper introduces a trainable neural network model for underwater image enhancement that improves colorfulness, contrast, and detail by combining preliminary color correction with self-adaptive refinement.

Contribution

A novel end-to-end neural enhancement model with a non-parameter layer and parametric layers for adaptive underwater image enhancement.

Findings

Achieves high-quality enhancement results on natural underwater scenes.

Joint optimization improves colorfulness, contrast, and detail.

Outperforms existing methods in visual quality metrics.

Abstract

Due to the absorption and scattering effects of the water, underwater images tend to suffer from many severe problems, such as low contrast, grayed out colors and blurring content. To improve the visual quality of underwater images, we proposed a novel enhancement model, which is a trainable end-to-end neural model. Two parts constitute the overall model. The first one is a non-parameter layer for the preliminary color correction, then the second part is consisted of parametric layers for a self-adaptive refinement, namely the channel-wise linear shift. For better details, contrast and colorfulness, this enhancement network is jointly optimized by the pixel-level and characteristiclevel training criteria. Through extensive experiments on natural underwater scenes, we show that the proposed method can get high quality enhancement results.