Image Magnification Using Correlation-Based Interpolation Coefficients
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image magnification devices face challenges in performing optimal interpolation computations due to edge shape detection methods that rely on comparison with predetermined patterns, leading to suboptimal results when edge shapes deviate slightly from these patterns.
Innovation Solution
An image magnification device and method that calculates interpolation coefficients based on the strength of correlation between pixels in a low-resolution image, using variation quantities and direction indication data to determine strong correlation directions, allowing for accurate interpolation computations without relying on predetermined edge shape patterns.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If edge shape detection is performed by comparison with predetermined patterns, then the detection process is simple and fast, but the detection accuracy deteriorates when edge shapes deviate from the patterns
Solution Approach 1:
The patent changes the parameter of edge shape representation from discrete predetermined patterns to continuous gradient magnitude and direction values. By calculating gradient magnitude and direction at each pixel and using these continuous parameters to determine interpolation coefficients, the system achieves both speed and accuracy in edge shape detection without being limited to predefined patterns.
2Measurement precision
If the number of comparison patterns is increased to improve edge shape detection sensitivity, then the detection accuracy improves, but the device complexity increases
Solution Approach 1:
The patent extracts the essential characteristics of edge shapes (gradient magnitude and direction) from the complex task of pattern matching. By taking out only the necessary gradient information and using it directly to compute interpolation coefficients, the system achieves high detection sensitivity without needing to store or compare multiple predetermined patterns, thus reducing device complexity.
3Ease of manufacture
If interpolation coefficients are determined using predetermined edge shape patterns, then the computation is straightforward, but the interpolation quality deteriorates for edges not matching the patterns
Solution Approach 1:
The patent changes the parameters used for determining interpolation coefficients from discrete pattern matches to continuous gradient magnitude and direction values. This allows the system to compute interpolation coefficients that accurately reflect the actual edge characteristics at each pixel, improving interpolation quality for edges of any shape while maintaining computational simplicity through direct gradient-based calculations.
Data Source
AI summary
An interpolation computation unit (3B) treats, as positions of interest, positions where pixels within a high-resolution image (D30) occupy when the high-resolution image (D30) is superimposed on a low-resolution image (D01), and for each position of interest, obtains a pixel value for a pixel assumed to exist at the position of interest by performing an interpolation computation using pixel values of a plurality of pixels within the low-resolution image (D01). An interpolation coefficient calculation unit (3A) obtains interpolation coefficients (D3A) having values that increase with increasing strength of correlation of the pixels in the plurality of pixels in the low-resolution image with the pixel of interest, and outputs the interpolation coefficients to the interpolation computation unit (3B). Angles of edges and shapes of edges are not classified into any predetermined patterns; therefore, it is possible to perform suitable interpolation computations regardless of edge shape.