Image Processing Apparatus Adaptive Filtering Resizing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing techniques that resize images captured by cameras with Bayer arrays face challenges in maintaining image quality due to differences in sampling structures between color filters, leading to issues like lowered perceived resolution and aliasing, and require white balance adjustments that can result in signal clipping and reduced flexibility in image quality adjustments.
Innovation Solution
An image processing apparatus that performs adaptive filtering and resizing by applying multiple white balance processing steps to generate color signals with equivalent ratios, allowing for interpolation and resizing while preventing signal clipping and maintaining image quality, using a processor-based system with a memory device to manage the processing units and circuits for color separation, filtering, and resizing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If white balance adjustment is performed in advance before resizing, then correlation detection for adaptive interpolation can be achieved, but signal clipping occurs and degree of freedom in image quality adjustment is reduced
Solution Approach 1:
The patent applies preliminary action by performing white balance adjustment on a copy of the input image rather than the original, allowing correlation detection to benefit from white balance processing while preserving the original signal for subsequent flexible adjustments. This resolves the contradiction by preparing necessary data in advance without compromising future adaptability.
Solution Approach 2:
The patent segments the image processing into distinct stages: creating a copy for white balance adjustment and correlation detection, while preserving the original input image for subsequent resizing and interpolation operations. This segmentation allows each stage to operate independently without interfering with the other, maintaining both measurement precision and adjustment flexibility.
2Productivity
If simple linear interpolation is used for resizing, then processing speed is improved, but image quality deteriorates due to aliasing and lowered perceived resolution
Solution Approach 1:
The patent changes the interpolation parameter from simple linear interpolation to adaptive interpolation that utilizes correlation detection results. The interpolation method dynamically adjusts based on local image characteristics detected through correlation analysis, improving image quality while maintaining reasonable processing speed through efficient correlation-based calculations.
Solution Approach 2:
The patent implements feedback by using correlation detection results to guide the interpolation process. The correlation information about pixel relationships is fed back into the interpolation algorithm, allowing it to adapt to local image structures and reduce aliasing artifacts while preserving perceived resolution.
3Manufacturing precision
If adaptive interpolation with correlation detection is performed, then image quality is maintained during resizing, but white balance adjustment causes signal clipping
Solution Approach 1:
The patent performs white balance adjustment on a separate copy of the input image before resizing operations. This preliminary action on a copy allows correlation detection to utilize white balance corrected data while the original unadjusted image remains available for resizing, preventing signal clipping in the final output.
Solution Approach 2:
The patent introduces an intermediary copy of the image that serves as a mediator between white balance adjustment and resizing operations. This intermediary contains the white balance adjusted data needed for correlation detection, while the original image serves as the source for resizing, preventing direct conflict between these two processing steps.
Data Source
AI summary
An image processing apparatus has: a WB processing circuit that performs a plurality of kinds of WB processing for an input image that is undeveloped and captured by an image capturing element in which pixels of a plurality of colors are arranged in a mosaic pattern under a predetermined rule; filter circuits that perform color interpolation processing by referring to the input image and a signal subjected to WB processing and generate images of respective color signals whose color ratio is equivalent to that of the input image; and resizing circuits that resize the generated images.