HDR Image Composition Using Motion Maps to Cut Blur and Noise
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Wide dynamic range (WDR) and high dynamic range (HDR) shooting methods cause motion blur and increase noise due to differences in exposure times, especially in scenes with high contrast ratios.
Innovation Solution
An image processing method that generates motion data using motion masks of varying sizes, creates motion maps based on these data, determines a reference motion map, and adjusts channel composition ratios to minimize motion artifacts and noise in the output image.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Illumination intensity
If long exposure images are used to expand dynamic range, then the dynamic range of the output image is improved, but motion blur and noise increase
Solution Approach 1:
The patent applies different processing strategies to different regions of the image based on motion detection. Static regions use long exposure images to maximize dynamic range, while moving regions use short exposure images to avoid motion blur. This local differentiation resolves the contradiction by allowing each region to have the exposure characteristics most suitable for its content.
Solution Approach 2:
The image is segmented into multiple channels with different exposure times, and motion regions are further segmented from static regions. This segmentation allows independent processing of different regions, enabling the system to combine the benefits of long exposure (high dynamic range) and short exposure (low motion blur) in different parts of the same image.
2Object-affected harmful factors
If short exposure time is used to reduce motion blur, then motion artifacts are reduced, but noise increases due to high digital gain
Solution Approach 1:
The patent merges multiple image channels with different exposure times and noise characteristics. By combining short exposure images (low motion artifacts, high noise) with long exposure images (high dynamic range, high motion blur), the system achieves a result that has neither severe motion artifacts nor excessive noise, as the complementary strengths of each channel offset their respective weaknesses.
Solution Approach 2:
The patent changes the exposure time parameter across different image channels, creating a set of images with varying exposure durations. This parameter variation allows the system to select or combine images based on local motion characteristics, optimizing the trade-off between motion artifacts and noise on a region-by-region basis.
3Illumination intensity
If multiple channels with different exposures are synthesized, then dynamic range is expanded, but image complexity and processing requirements increase
Solution Approach 1:
The patent performs preliminary motion detection and region classification before the actual image synthesis. By pre-identifying motion regions and assigning appropriate exposure channels to each region, the system simplifies the subsequent synthesis process and avoids the need for complex real-time optimization during image merging.
Data Source
AI summary
Provided is an image processing method including generating motion data for a pixel included in each of one or more channels configured to use motion masks having different sizes, the channels being included in at least a portion of an output image, generating one or more motion maps based on a difference between motion data the one or more channels for motion masks having a same size, determining a reference motion map among the one or more motion maps for each pixel included in the output image, determining a composition ratio of the one or more channels to be used for each pixel included in the output image based on a motion value on the reference motion map, and generating the output image based on the composition ratio of the one or more channels.