Naturalness Preserved Image Enhancement Using a priori Multi-Layer Lightness Statistics

Area of Science:

• Computer Vision and Image Processing

Background:

• Non-uniform illumination in images often leads to over-enhancement and unnatural visual results.
• Existing image enhancement methods struggle to balance contrast improvement with naturalness preservation.

Purpose of the Study:

• To develop a novel image enhancement method that preserves naturalness in non-uniformly illuminated images.
• To incorporate prior multi-layer lightness statistics from high-quality images into an enhancement model.

Main Methods:

• A novel multi-layer image enhancement model was designed.
• Multi-layer lightness statistics were derived from high-quality outdoor images.
• Human observer studies evaluated naturalness preservation and overall image quality.

Main Results:

• The proposed method demonstrated superior performance compared to four state-of-the-art enhancement techniques.
• Enhanced image quality ratings correlated with both contrast enhancement and naturalness preservation.
• Human evaluations confirmed the method's effectiveness in preserving naturalness.

Conclusions:

• The developed method effectively enhances non-uniformly illuminated images while maintaining a high degree of naturalness.
• The integration of multi-layer lightness statistics provides a robust approach to image enhancement.
• The findings suggest a new direction for developing more visually pleasing image enhancement algorithms.