Single-image dehazing method based on Rayleigh Scattering and adaptive color compensation

Area of Science:

• Computer Vision
• Image Processing
• Remote Sensing

Background:

• Degradation in outdoor images due to atmospheric light and scattering is a common challenge.
• Existing methods like Dark Channel Prior (DCP) can fail in certain regions, leading to inaccurate transmission estimation and image artifacts.
• Color dilution and loss of detail are significant issues in dehazing algorithms.

Purpose of the Study:

• To develop an improved image dehazing method that addresses the limitations of existing techniques.
• To enhance image quality by accurately estimating and compensating for atmospheric light and color.
• To effectively segment and repair regions where conventional methods fail.

Main Methods:

• A novel method combining Rayleigh Scattering compensation and adaptive color correction.
• B-channel compensation for atmospheric illumination, iteratively refined for accuracy.
• Joint evaluation of dark and bright channel images to identify and re-estimate transmission in failure regions.
• Regional segmentation based on brightness and color differentials.

Main Results:

• Elimination of color dilution through iterative B-channel compensation.
• Improved image quality and detail restoration in previously problematic areas.
• Demonstrated effectiveness and resilience across diverse experimental scenarios and parameter settings.

Conclusions:

• The proposed method significantly enhances image dehazing performance.
• It offers a robust solution for improving image quality in challenging outdoor environments.
• The technique effectively overcomes limitations of the Dark Channel Prior method.