Fog Removal in Images Using Airlight Estimation and Anisotropic Diffusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for fog removal from images and videos are either ineffective in dense fog conditions, require multiple images, or involve complex user-dependent processes, failing to achieve high perceptual quality with reduced noise and enhanced contrast in real-time applications.
Innovation Solution
A method and system that utilize airlight estimation and refinement through anisotropic diffusion or bilateral filtering to restore foggy images and videos, allowing for single-camera fog removal with reduced computational requirements, effective in both RGB and gray scale models, and adaptable for real-time video encoding or decoding.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple images are used for fog removal, then depth map estimation accuracy is improved, but processing time and complexity increase
Solution Approach 1:
The patent performs histogram equalization on the foggy image before airlight estimation to enhance contrast and improve the accuracy of subsequent depth map estimation. This preliminary processing step prepares the image data to facilitate more accurate fog removal without requiring multiple input images
Solution Approach 2:
The patent introduces an airlight map as an intermediary representation that captures the spatial distribution of atmospheric light. This airlight map serves as a mediator between the foggy input image and the final restored image, enabling accurate depth estimation and fog removal through refined airlight estimation using anisotropic diffusion or bilateral filtering
2Manufacturing precision
If complex filtering methods are used for airlight estimation, then fog removal quality is improved, but computational cost increases
Solution Approach 1:
The patent applies anisotropic diffusion filtering that adapts the filtering strength locally based on image gradients. Regions with strong edges maintain sharp boundaries while regions with smooth gradients undergo stronger smoothing, achieving high-quality airlight estimation without uniformly applying heavy computational filtering across the entire image
Solution Approach 2:
The patent performs histogram equalization followed by airlight estimation with optional refinement filtering. The refinement step using anisotropic diffusion or bilateral filtering is applied selectively to enhance airlight map quality where needed, rather than applying full-strength filtering uniformly, thus achieving high fog removal quality with optimized computational cost
3Illumination intensity
If histogram equalization is applied to foggy images, then contrast is enhanced, but noise may be amplified
Solution Approach 1:
Histogram equalization is applied as a preliminary step before airlight estimation to enhance the contrast of the foggy image. This improves the visibility of structural features and facilitates more accurate atmospheric light estimation in subsequent processing steps
Solution Approach 2:
The airlight map acts as an intermediary that separates the atmospheric scattering component from the scene radiance. By estimating and removing the airlight component through the transmission map, the patent effectively reduces the amplified noise while preserving the enhanced contrast benefits from histogram equalization
Data Source
AI summary
A method of removing fog from the images/videos independent of the density or amount of the fog and free of user intervention and a system for carrying out such method of fog removal from images/videos are disclosed. The removal of fog from images and video involve airlight estimation and airlight map refinement based restoration of foggy images and videos. Advantageously, removal of fog from images and videos of this invention would require less execution time and yet achieve high perceptual image quality with reduced noise and enhanced contrast. The proposed method is adapted for RGB Color model and advantageously also for HSI color model involving reduced computational requirements and be user friendly and supposed to have wide application and use.