Optical Flow Haze Removal Iterative Refinement
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional optical flow algorithms fail to account for haze in images, leading to inaccurate motion estimation and object recognition in scenes obscured by atmospheric particles like fog and smoke, as they assume consistent color and contrast across images, which is not the case due to changing haze conditions.
Innovation Solution
An iterative process combining image de-hazing and optical flow computation, where the haze is initially cleared from digital images based on an estimate of light contributed by the haze, allowing for improved visibility and recognition of objects across images, and refining the images based on computed optical flow to maintain temporal consistency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional optical flow algorithms are used, then processing speed is maintained, but motion estimation accuracy deteriorates due to haze interference
Solution Approach 1:
The patent applies preliminary action by performing haze removal as a preprocessing step before computing optical flow. The haze is estimated and removed from images prior to motion estimation, ensuring that the optical flow algorithm operates on cleared images with consistent color and contrast, thereby improving motion estimation accuracy without adding complexity during the core processing stages.
Solution Approach 2:
The patent introduces an intermediary haze removal module between the image capture and optical flow computation stages. This intermediary component estimates and removes haze effects, serving as a mediator that prepares the images for accurate optical flow analysis while maintaining the separation of concerns between haze removal and motion estimation algorithms.
2Illumination intensity
If single image de-hazing is applied, then image visibility is improved, but temporal consistency across image sequences deteriorates
Solution Approach 1:
The patent implements feedback by using the optical flow information to guide and refine the haze removal process. The system iteratively adjusts the haze estimation and removal based on the computed optical flow, ensuring that the de-hazed images maintain temporal consistency across sequences while preserving improved visibility. The feedback loop ensures that haze removal respects the temporal structure of the image sequence.
Solution Approach 2:
The patent merges the haze removal process with the optical flow computation into a unified iterative framework. By combining these two previously separate operations, the system achieves both improved image visibility and temporal consistency, as the merged process considers both haze effects and motion information simultaneously across the image sequence.
3Manufacturing precision
If haze removal algorithms are applied to image sequences, then image quality is improved, but computational time increases
Solution Approach 1:
The patent applies preliminary action by performing haze removal as a preprocessing step before optical flow computation. This staged approach allows the system to clear haze from images beforehand, improving image quality for subsequent processing while managing computational time through efficient sequential execution rather than simultaneous processing of all operations.
Solution Approach 2:
The patent applies partial action by removing haze selectively based on the estimated haze distribution in each image region. Rather than uniformly processing entire images, the system applies haze removal algorithms only to regions where haze is detected, reducing unnecessary computational operations while maintaining image quality where needed.
Data Source
AI summary
In embodiments of optical flow accounting for image haze, digital images may include objects that are at least partially obscured by a haze that is visible in the digital images, and an estimate of light that is contributed by the haze in the digital images can be determined. The haze can be cleared from the digital images based on the estimate of the light that is contributed by the haze, and clearer digital images can be generated. An optical flow between the clearer digital images can then be computed, and the clearer digital images refined based on the optical flow to further clear the haze from the images in an iterative process to improve visibility of the objects in the digital images.