Scene Segmentation-Based Luminance Adjustment for Multi-Exposure Image Fusion.

Area of Science:

• Computer Vision
• Image Processing

Background:

• Multi-exposure image fusion combines photos with varying exposures to create superior images.
• Existing methods struggle with insufficient exposure variation, leading to unclear fused images.

Purpose of the Study:

• To propose a novel method for adjusting luminance in multi-exposure image fusion.
• To enhance the quality of fused images, particularly with challenging input data.

Main Methods:

• A novel luminance adjustment technique is proposed for multi-exposure image fusion.
• Two new scene segmentation approaches based on luminance distribution are introduced for adjustment.
• The method improves fused image quality by adjusting input image luminance.

Main Results:

• The proposed method produces high-quality, clear fused images, even from undesirable inputs.
• Visual comparisons demonstrate superior scene representation compared to existing methods.
• Quantitative metrics (MEF-SSIM, entropy, TMIQI, naturalness) show improved performance over state-of-the-art techniques.

Conclusions:

• The novel luminance adjustment method significantly enhances multi-exposure image fusion.
• The approach overcomes limitations of existing methods, producing clearer and more visually appealing fused images.
• The proposed technique offers a robust solution for generating high-quality fused images across various input conditions.