Foggy Image Restoration via Depth Map and Optical Model
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Fog in images captured by outdoor cameras degrades image quality due to uniform and continuous visibility reduction, leading to reduced contrast and chroma, making it difficult to restore clear images.
Innovation Solution
An image processing method that generates a pixel depth image by estimating pixel depths based on channel differences in foggy images, processes the pixel depth image through filtering and refinement, and uses optical model parameters to restore images as if they were taken without fog, involving depth estimation, gray-scale transformation, and calculation of scattering coefficients.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If fog particles scatter light in the air, then visibility is reduced and image quality is degraded, but contrast and chroma are uniformly reduced across the entire image
Solution Approach 1:
The patent segments the image processing into multiple stages: depth map generation, atmospheric light estimation, and pixel-by-pixel restoration. By dividing the foggy image into regions with different depth values and applying different restoration parameters to each region, the method achieves selective recovery of visibility while preserving image quality.
Solution Approach 2:
The patent changes physical parameters by estimating depth values and atmospheric light properties for each pixel, then uses these parameters to calculate transmission maps and apply targeted restoration. This parameter-based approach allows the system to adapt to varying fog conditions across different regions of the image.
2Productivity
If conventional image processing methods are applied to foggy images, then processing time is reduced, but restoration accuracy and image quality are insufficient
Solution Approach 1:
The patent performs preliminary actions by first generating a depth map and estimating atmospheric light before the actual restoration process. These preliminary steps provide crucial information about scene geometry and lighting conditions, enabling more accurate and efficient restoration in subsequent processing stages.
Solution Approach 2:
The patent introduces intermediate representations including depth maps, transmission maps, and atmospheric light estimates as mediators between the input foggy image and the final restored image. These intermediaries carry essential information through the processing pipeline, enabling accurate restoration while maintaining computational efficiency.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
The method effectively improves image contrast and color sense by restoring images degraded by fog, enhancing visibility and clarity.
Implementation Method 1
Visibility degradation due to fog is caused by collisions of the fog particles in the air with light and scattering of the fog particles. Light generated due to such scattering is referred to as airlight.
Implementation Method 2
obtaining a scattering coefficient from an image which is gray-scaled-transformed from the foggy image
Data Source
AI summary
An image processing apparatus and method for restoring an image which is expected to be when there is no fog from a foggy image, the image processing method including: generating a pixel depth image of the foggy image by estimating a plurality of pixel depths of a plurality of pixels, respectively, included in the foggy image based on a channel difference between at least two of red (R), green (G), and blue (B) channels of the foggy image; processing the pixel depth image; obtaining an optical model parameter with respect to the foggy image; and restoring an image, which is expected to be when the foggy image does not include the fog, by using the processed pixel depth image and the optical model parameter.