Low Power Patch Matching for Single Frame Super-Resolution
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional super-resolution techniques for displays require significant computational resources and power, limiting their effectiveness on devices with limited resources such as smartphones and tablets, especially when larger magnification ratios or motion are involved, and struggle to maintain high image quality.
Innovation Solution
A low power patch matching method for single frame self-similarity based super-resolution that analyzes the search range and motion to determine if patch matching is necessary, using efficient techniques like bilinear interpolation and parabolic prediction to reduce computational requirements and conserve power.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional multi-image super resolution techniques are used to construct high resolution images, then image detail reconstruction is improved, but computational complexity and hardware resource requirements increase significantly
Solution Approach 1:
The patent segments the super-resolution process into two distinct stages: (1) an initial super-resolution pass that generates a high-resolution image using multi-image techniques, and (2) a refinement pass that performs selective patch matching only on regions requiring improvement. This segmentation reduces overall computational complexity while maintaining image detail quality.
Solution Approach 2:
The patent applies local quality by performing patch matching selectively only in regions where it is beneficial, rather than uniformly across the entire image. The system identifies regions with motion or large magnification ratios and applies computationally intensive patch matching only there, while using simpler interpolation methods in other regions, thus reducing hardware resource requirements.
2Measurement precision
If patch matching is performed to improve high resolution image quality, then image fidelity is improved, but computational requirements and power consumption increase
Solution Approach 1:
The patent applies partial action by performing patch matching selectively rather than universally. It performs full patch matching only in regions where it is necessary (areas with motion or large magnification), while using simpler methods elsewhere. This partial application maintains image fidelity in critical regions while significantly reducing power consumption compared to applying patch matching to the entire image.
Solution Approach 2:
The patent performs preliminary super-resolution processing to generate a high-resolution image before performing patch matching. This preliminary action provides a foundation that reduces the computational burden of subsequent patch matching operations, as the algorithm only needs to refine specific regions rather than generate the entire high-resolution image from scratch.
3Adaptability or versatility
If larger magnification ratios are applied in super resolution, then display resolution compatibility is improved, but image quality and detail reconstruction deteriorate
Solution Approach 1:
The patent applies dynamics by adaptively adjusting the patch matching search range based on the magnification ratio and motion characteristics of different image regions. For regions with large magnification ratios, the system dynamically expands the search range to find better matching patches, while using smaller search ranges in regions with smaller magnification ratios. This dynamic adjustment maintains image quality across varying magnification levels while ensuring compatibility with different display resolutions.
Data Source
AI summary
A method of generating a super-resolution image from a single frame of image data includes using a processor to retrieve query patches of image data from a memory, determining a search range for each patch, and generating super-resolution image data corresponding to each patch based upon the search range.