Image Interpolation Datum Point Shift for Hue Uniformity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image resolution conversion methods, such as bicubic and bilinear interpolation, often result in hue discrepancies between edge and inner pixels in the converted image, leading to undesirable color harmonization and conspicuous joints in tiling patterns.
Innovation Solution
An image processing apparatus and method that determines the coordinates of a datum point for pixel value interpolation by shifting the coordinates of end pixels in the converted image further inside the input image than the original end coordinates, ensuring harmonized hues across the image, particularly by adjusting the shift amount based on the position of intervening pixels.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the datum point is aligned with the position of the pixel of the input image, then the pixel value of the pixel corresponding to the datum point in the converted image becomes a pixel value of the pixel at the position of the datum point in the input image, but there is a difference in the hue between the pixels present on the two sides passing through the origin and the pixels present on the inner side in the converted image
Solution Approach 1:
The patent applies different datum point positioning strategies for different regions of the image. For pixels at the ends in the first direction, the datum point is positioned at coordinates shifted further inside the input image than the end coordinates. For other pixels, the datum point is positioned at coordinates corresponding to the pixel positions. This local differentiation resolves the hue uniformity issue at edges while maintaining pixel value accuracy for interior pixels.
Solution Approach 2:
The patent changes the coordinate parameter of the datum point based on the position of the target pixel. Specifically, it introduces a coordinate shift for end pixels in the first direction, where the datum point coordinates are adjusted to be further inside the input image. This parameter change eliminates the hue difference between edge and inner pixels while preserving the interpolation accuracy.
2Device complexity
If coordinates in the first direction of the datum point of the second pixels present at both ends in the first direction are determined to be coordinates aligned with the first direction end portion coordinates, then the interpolation operation is simplified, but the hue difference between edge and inner pixels becomes conspicuous
Solution Approach 1:
The patent introduces a local modification to the datum point coordinate determination. For pixels at the ends in the first direction, the datum point coordinates are shifted further inside the input image compared to the standard alignment. This localized coordinate adjustment maintains simplicity of the overall interpolation operation while specifically addressing the hue uniformity problem at image edges.
Solution Approach 2:
The patent modifies the coordinate parameter of the datum point for end pixels in the first direction. By determining the datum point coordinates to be shifted further inside the input image than the end coordinates, it creates a parameter change that eliminates conspicuous hue differences while keeping the interpolation operation straightforward for all other pixels.
3Ease of operation
If the same datum point alignment method is applied to all pixels, then the processing is consistent and simple, but the color harmonization is poor at image edges
Solution Approach 1:
The patent implements a conditional datum point positioning system. For pixels at the ends in the first direction, the datum point is positioned at coordinates shifted further inside the input image. For other pixels, the datum point aligns with the pixel position. This local differentiation maintains processing consistency through clear conditional logic while significantly improving color harmonization at image edges.
Data Source
AI summary
An image processing apparatus includes a datum point determination unit configured to determine coordinates of a datum point for determining pixel values of second pixels included in a converted image including a plurality of second pixels in a coordinate plane of an input image including a plurality of first pixels, and an interpolation unit configured to determine pixel values of the second pixels based on a plurality of reference pixels. The datum point determination unit sets coordinates in the first direction of the first pixels among the plurality of first pixels as first direction end portion coordinates, and determines coordinates in the first direction of the datum point of the second pixels among the plurality of second pixels to be coordinates shifted to further inside of the input image than the first direction end portion coordinates by a distance longer than 0 and shorter than one pixel.