Super Resolution Signal Processing via Self-Similarity Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing super-resolution methods face limitations in noise resilience and misregistration, particularly when attempting to achieve high super-resolution factors, and often require external databases for high-frequency information.
Innovation Solution
The method exploits data redundancy within and across different scales of an input signal to combine high-frequency information, using patch recurrence and transformations like rotation, reflection, and scaling to generate a super-resolution version without relying on external databases, by solving equations based on blur kernels and sub-pixel misalignments.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If external databases are used to provide high-frequency information for super-resolution, then super-resolution quality can be improved, but device complexity and data storage requirements increase
Solution Approach 1:
The patent applies self-service by extracting high-frequency information from the input signal itself through self-similarity analysis, rather than relying on external databases. The method analyzes redundant information within the input signal across different scales and orientations to reconstruct high-frequency details, making the system self-sufficient and eliminating the need for external data sources.
2Loss of information
If multiple low-resolution images are used for super-resolution, then high-frequency information can be recovered, but misregistration and noise issues worsen
Solution Approach 1:
The patent segments the super-resolution process into distinct stages: extracting low-frequency information from the input signal, analyzing self-similarity patterns across different scales and orientations, and reconstructing high-frequency information through mathematical relationships. This segmentation allows each stage to be optimized independently, reducing the cumulative effect of misregistration and noise.
Solution Approach 2:
The patent performs preliminary analysis of the input signal to identify self-similarity patterns and extract structural information before the actual super-resolution reconstruction. By pre-processing the signal to establish relationships between different scales and orientations, the method prepares the data in a form that is more resistant to misregistration and noise during the final reconstruction phase.
3Measurement precision
If high super-resolution factors are attempted, then resolution improvement increases, but noise and misregistration effects are amplified
Solution Approach 1:
The patent extends the analysis from a single scale to multiple scales and orientations, adding dimensional aspects to the super-resolution process. By examining the input signal at different scales (e.g., 1/2, 1/4, 1/8 of original size) and orientations (0°, 45°, 90°, 135°), the method captures high-frequency information from multiple dimensional perspectives, which helps suppress noise and misregistration effects through redundant information.
Data Source
AI summary
A method implementable on a computing device includes exploiting data redundancy to combine high frequency information from at least two different scales of an input signal to generate a super resolution version of said input signal. An alternative method includes exploiting recurrence of data from an input signal in at least two different scales of at least one reference signal to extract and to combine high frequency information from a plurality of scales of said at least one reference signal to generate a super resolution version of said input signal. An alternative method includes generating a super resolution version of a single input video sequence in at least the temporal dimension by exploiting data recurrence within the input video sequence or with respect to an external database of example video sequences. A signal may be an image, a video sequence, an audio signal, etc.