Image Processing Device for Knurl Artifact Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing techniques face challenges in accurately positioning image capturing elements for high-resolution image combination, leading to alignment errors and artifacts like knurl artifacts, especially at edges with large luminance differences, due to limitations in infinite-resolution image spaces and motion estimation errors.
Innovation Solution
An image processing device and method that aligns and combines images in both high-resolution and low-resolution spaces based on displacement, calculates feature quantities for pixel-value changes, and applies correction filtering to reduce alignment errors and artifacts, using a combination of high-resolution combining, low-resolution weighted addition, and feature-quantity calculation units.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If pixels are arranged in a high-resolution image space to produce a high-resolution combined image, then the resolution of the combined image is improved, but alignment errors occur due to the finite resolution of the image space, causing knurl artifacts
Solution Approach 1:
The patent introduces a new dimension by calculating feature quantities (edge directions) at multiple orientations (0°, 45°, 90°, 135°) and using these multi-dimensional features to determine combining ratios. This allows the system to operate effectively in a finite-resolution image space while achieving alignment accuracy equivalent to infinite-resolution space by leveraging directional information from multiple angles.
Solution Approach 2:
The patent changes the parameter used for alignment from direct pixel position matching to feature quantity-based matching using edge directions at multiple orientations. By transforming the alignment criterion from spatial coordinates to directional features, the system can achieve higher effective precision in the finite-resolution image space without suffering from knurl artifacts.
2Measurement precision
If the image capturing element is highly accurately positioned and held stationary during exposure, then the resolution-increasing effect is improved, but positioning and stabilization difficulty increases
Solution Approach 1:
The patent employs feedback by calculating feature quantities from the captured images and using these features to determine optimal combining ratios. The system continuously adjusts the combining process based on the extracted edge direction features, allowing it to compensate for positioning errors and motion without requiring highly accurate initial positioning or stabilization.
Solution Approach 2:
The patent performs preliminary action by calculating feature quantities and determining combining ratios before final image synthesis. By pre-processing the images to extract directional features and determine optimal weighting, the system can later combine images effectively even if positioning was not perfectly accurate, thereby reducing the stringency of positioning and stabilization requirements.
3Manufacturing precision
If an infinite-resolution image space is used to arrange pixels without alignment error, then alignment accuracy is improved, but memory requirements become impractically large
Solution Approach 1:
The patent changes the fundamental parameter for alignment from spatial coordinates in an infinite-resolution space to feature quantities (edge directions) calculated at multiple orientations. This parameter transformation allows the system to achieve equivalent alignment accuracy using a finite-resolution image space that fits within available memory, by operating in the feature quantity domain rather than the raw pixel domain.
Solution Approach 2:
The patent introduces feature quantities (edge directions at 0°, 45°, 90°, 135°) as an intermediary between the captured images and the final combined image. These feature quantities serve as a mediator that enables accurate alignment and combining ratio determination in a finite-resolution space, avoiding the need for an infinite-resolution image space while maintaining high alignment accuracy.
4Manufacturing precision
If correction processing is applied to regions with large alignment error, then knurl artifacts are reduced, but processing complexity increases
Solution Approach 1:
The patent applies local quality by determining different combining ratios for different regions of the image based on locally calculated feature quantities. Instead of applying uniform correction processing across the entire image, the system calculates edge directions locally and uses these local features to determine region-specific combining ratios, thereby reducing knurl artifacts where needed while maintaining simplicity in regions where correction is less critical.
Solution Approach 2:
The patent applies partial action by focusing correction processing only on regions where it is most needed - areas with large alignment errors and significant knurl artifacts. By using feature quantities to identify and prioritize correction in problematic regions rather than applying uniform processing across the entire image, the system achieves effective artifact reduction with reduced overall processing complexity.
Data Source
AI summary
An image processing device comprises: one or more processors comprising hardware, the one or more processors being configured to: generate a high-resolution combined image by aligning a plurality of time-series images with each other in a high-resolution image space having a resolution higher than the plurality of time-series images based on an amount of displacement between the plurality of time-series images, and combining the plurality of time-series images; generate a low-resolution combined image by aligning the plurality of time-series images with each other in a low-resolution image space having a resolution equal to or lower than the resolution of the plurality of time-series images based on the amount of displacement, combining the plurality of time-series images through weighted addition; calculate a feature quantity pertaining to a pixel-value change direction at each region in the generated low-resolution combined image; and correct the high-resolution combined image based on the calculated feature quantity.