Image Deblurring Using Dynamic Vision Sensor Events
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image deblurring methods using deconvolution are ineffective when the cause of blurring is unknown, as they rely on predetermined target values and may not accurately restore sharpness in all cases.
Innovation Solution
An image deblurring method that utilizes a dynamic vision sensor (DVS) event set to estimate a DVS edge estimation image, aligns and overlays images across time slices to determine camera motion, and applies deconvolution transform based on the alignment results to deblur images effectively.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If deconvolution transform is used for deblurring, then the deblurring process can be implemented, but the method is ineffective when the cause of blurring is unknown
Solution Approach 1:
The patent introduces a DVS event set as an intermediary element that captures camera motion information independently of the blurred image content. This mediator provides motion data that guides the deconvolution process, making the method effective even when the blurring cause is unknown. The DVS events serve as a bridge between the blurred image and the restoration process, enabling reliable deblurring without requiring prior knowledge of motion patterns.
2Ease of manufacture
If predetermined target values are set for deblurring, then the deblurring method can be executed, but it cannot accurately restore sharpness when actual images do not satisfy the set values
Solution Approach 1:
The patent transforms the static predetermined target values into dynamic, image-specific parameters by using DVS event sets to capture actual camera motion during exposure. The deblurring parameters are adapted based on the specific motion patterns detected in each image, allowing the method to maintain ease of implementation while achieving high accuracy in sharpness restoration for diverse imaging conditions.
3Measurement precision
If images from multiple time slices are aligned and overlaid, then camera motion can be accurately estimated, but the processing complexity increases
Solution Approach 1:
The patent segments the exposure time into multiple time slices and processes DVS events from each slice separately before alignment and overlay. This segmentation approach enables accurate camera motion estimation by capturing motion at different temporal points, while the modular processing structure manages complexity through systematic organization of the alignment and overlay operations.
Data Source
AI summary
An image deblurring method and an image deblurring apparatus are provided. The image deblurring method includes acquiring a blurred image and a dynamic vision sensor (DVS) event set that is recorded by a DVS while the blurred image is exposed, and deblurring the blurred image, based on the DVS event set.