Perceptual illumination-structure patch decomposition for enhancing complex lighted underwater images.

Area of Science:

• Computer Vision
• Image Processing
• Underwater Imaging

Background:

• Underwater images are degraded by complex lighting, causing uneven exposure, color cast, low contrast, and blurred details.
• Existing enhancement methods often fail to adequately brighten dark regions or suppress overexposed areas.

Purpose of the Study:

• To propose a novel model for enhancing complex lighted underwater images.
• To address the limitations of existing methods in handling uneven exposure and preserving details.

Main Methods:

• A perceptual illumination-structure patch decomposition (PISPD) model is proposed.
• The PISPD method utilizes complementary inputs: a contrast-enhanced image and a detail-sharpened image.
• It decomposes inputs into a perceptual illumination map, contrast, structure, and average intensity, incorporating a weighted edge-preserving factor.

Main Results:

• The PISPD model effectively balances brightness and integrates features from complementary inputs.
• A new benchmark dataset, CLUID, comprising 990 complex lighted underwater images, is introduced.
• Experimental results demonstrate superior performance of PISPD over eight state-of-the-art approaches.

Conclusions:

• The PISPD model offers a significant advancement in enhancing complex lighted underwater images.
• The method successfully mitigates common underwater image degradations like uneven exposure and low contrast.
• The introduced CLUID dataset facilitates further research in underwater image enhancement.