Image De-hazing via Atmospheric Light Estimation and Bilateral Filtering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing de-hazing methods for images are computationally intensive and often rely on dark channel priors, which are not efficient for real-time processing.
Innovation Solution
The method estimates global atmospheric light using k-means clustering and initial transmission values, applying a solver or bilateral filtering to recover scene radiance and construct haze-free images without relying on dark channel priors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If contrast enhancement techniques are used to counter haze effects, then image quality is improved, but computational time increases to approximately 10 seconds
Solution Approach 1:
The patent replaces expensive, computationally intensive contrast enhancement techniques with a simpler, faster alternative based on dark channel prior. This 'cheap' method achieves comparable image quality results in significantly less time, sacrificing the complexity of previous methods for speed and efficiency.
Solution Approach 2:
The patent changes the fundamental parameter approach by introducing the dark channel prior concept - observing that most local patches in haze-free images contain pixels with very low intensities in at least one color channel. This parameter observation enables a completely different, more efficient computational pathway that avoids the time-consuming contrast enhancement processes.
2Productivity
If dark channel prior is used to estimate haze thickness, then de-hazing is achieved faster, but the method becomes more complex compared to simple contrast enhancement
Solution Approach 1:
The dark channel prior method is self-service in that it automatically identifies and exploits the inherent property of haze-free images (dark channels) without requiring complex external processing. The method serves itself by using the image's own statistical properties to guide the de-hazing process, achieving both speed and simplicity.
3Length of moving object
If object distance from camera increases, then scene depth is captured, but haze blending with object color becomes more severe
Solution Approach 1:
The patent applies local quality by operating on local patches rather than the entire image uniformly. By analyzing dark channel properties in localized regions, the method can adaptively handle varying haze conditions at different depths, preserving scene depth information while effectively removing haze blending effects in each local area.
Data Source
AI summary
An image processing server performs haze-removal from images. Global atmospheric light is estimated and an initial transmission value is estimated. In one embodiment, a solver is applied to an objective function to recover a scene radiance value based on the estimated atmospheric light and estimated transmission value. The scene radiance value is used to construct an image without haze. In a simplified method that avoids using a solver, bilateral filtering is performed on the transmission image in order to construct an image without haze.