Fast single image haze removal via local atmospheric light veil estimation

Area of Science:

• Computer Vision
• Image Processing

Background:

• Haze significantly degrades image quality and obscures details.
• Existing dehazing methods often struggle with accurate global atmospheric light estimation, leading to artifacts.

Purpose of the Study:

• To propose a novel single-image dehazing framework.
• To enhance image visual quality and color fidelity by addressing haze artifacts.

Main Methods:

• A physical model-based approach utilizing local atmospheric light estimation.
• Employing extreme pixel intensity to define an initial atmospheric veil.
• Applying an across bilateral filter for local smoothness and edge preservation.
• Constructing transmission maps and reflection components from a physical atmospheric scattering model.

Main Results:

• The proposed method effectively removes haze from outdoor images.
• Achieved satisfactory visual quality and superior color fidelity in dehazed images.
• Demonstrated higher performance compared to several state-of-the-art dehazing methods.

Conclusions:

• The novel framework provides an effective solution for single-image dehazing.
• Local atmospheric light estimation overcomes limitations of global estimation.
• The method offers improved performance and visual output for hazy image restoration.