UIEC^2-Net: CNN-based Underwater Image Enhancement Using Two Color Space

Abstract: Underwater image enhancement has attracted much attention due to the rise of marine resource development in recent years. Benefit from the powerful representation capabilities of Convolution Neural Networks(CNNs), multiple underwater image enhancement algorithms based on CNNs have been proposed in the last few years. However, almost all of these algorithms employ RGB color space setting, which is insensitive to image properties such as luminance and saturation. To address this problem, we proposed Underwater Image Enhancement Convolution Neural Network using 2 Color Space (UICE^2-Net) that efficiently and effectively integrate both RGB Color Space and HSV Color Space in one single CNN. To our best knowledge, this method is the first to use HSV color space for underwater image enhancement based on deep learning. UIEC^2-Net is an end-to-end trainable network, consisting of three blocks as follow: a RGB pixel-level block implements fundamental operations such as denoising and removing color cast, a HSV global-adjust block for globally adjusting underwater image luminance, color and saturation by adopting a novel neural curve layer, and an attention map block for combining the advantages of RGB and HSV block output images by distributing weight to each pixel. Experimental results on synthetic and real-world underwater images show the good performance of our proposed method in both subjective comparisons and objective metrics. The code are available at https://github.com/BIGWangYuDong/UWEnhancement.

• The paper introduces a CNN model that integrates both RGB and HSV color spaces to address underwater image challenges.
• UIEC2-Net employs three core blocks—a pixel-level RGB block, a global-adjust HSV block, and an attention map block—to enhance image quality.
• Experimental results demonstrate superior performance over existing methods, with improved metrics such as MSE, PSNR, and SSIM.

Enhancing Underwater Imagery Through Dual Color Space Deep Learning: An Examination of UIEC^{}2-Net

The paper "UIEC^{}2-Net: CNN-based Underwater Image Enhancement Using Two Color Space" introduces a novel approach to improving the quality of underwater images by leveraging the advantages of convolutional neural networks (CNNs) in combination with RGB and HSV color spaces. This research responds to the significant challenges posed by underwater imagery, which is often plagued by issues such as poor contrast, color deviations, and luminance disparities due to the complex light interactions under water.

Overview of the Method

The proposed method, UIEC^{}2-Net, enhances underwater images by integrating operations in both the RGB and HSV color spaces within a single end-to-end trainable CNN model. The architecture is structured into three core blocks:

1. RGB Pixel-Level Block: This initial stage handles fundamental processes such as denoising and removing color casts using a straightforward CNN without the complexities of skip connections or downsampling, which preserves pixel-level details.
2. HSV Global-Adjust Block: It introduces global adjustments by employing a neural curve layer over HSV-transformed images, focusing on altering luminance and saturation, aiming to refine image properties from a global perspective. Notably, this is reported as the first instance of HSV color space utilization within a deep learning framework for underwater image processing.
3. Attention Map Block: This component integrates the outputs from RGB and HSV blocks, applying attention-based weighting at each pixel to amalgamate the strengths of both outputs while mitigating respective weaknesses.

Experimental Validation

The paper documents thorough experimentation on both synthetic and real-world underwater images, highlighting the efficacy of UIEC^{}2-Net compared to existing methods. The results included subjective visual assessments and objective measures such as MSE, PSNR, and SSIM. The proposed model demonstrated superior performance, producing results that exhibit less color distortion and better retention of image details. Enhanced visual quality was achieved across various underwater scenes, including those with pronounced blue and green hues, as well as scenes captured in low-light conditions.

Implications and Future Directions

By addressing the limitations of prior RGB-centric approaches, this research posits that incorporating HSV space significantly enhances an algorithm's ability to rectify underwater image issues, particularly those related to color and brightness which are not adequately addressed by RGB alone. This dual space approach allows UIEC^{}2-Net to outperform its predecessors both in terms of perceptual quality and quantitative metrics.

The practical implications of this research are far-reaching in areas demanding high-quality underwater imagery, such as marine biology, archaeology, and underwater exploration. The methodology sets a precedent for future work that could further explore variations and enhancements in color space integration within CNN frameworks, potentially leading to more generalized solutions applicable across different types of degraded image environments.

In conclusion, UIEC^{}2-Net exemplifies a significant methodological advancement in underwater image enhancement, offering a robust tool that improves on existing technologies by leveraging dual color spaces, and it stands as a proposal for subsequent advancements and standard practices in underwater image processing.