Super-Resolution Image Prediction Using Reliability-Weighted Geometric Transformation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The method described in existing techniques for generating high-resolution images from low-resolution images experiences a decrease in image quality due to failures in image registration during video super-resolution processing.
Innovation Solution
An image processing system that computes geometric transformation parameters and reliability levels to predict and generate super-resolution images by simulating a low-resolution image from a previous frame, calculating differences, and performing weighted mean calculations based on the reliability of the transformation parameters.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If image registration is performed to generate super-resolution images from multiple low-resolution images, then the resolution of the generated image is improved, but the image quality deteriorates when registration fails
Solution Approach 1:
The patent evaluates the reliability of geometric transformation parameters before using them to generate super-resolution images. By performing this reliability evaluation in advance, the system can determine whether to trust the registration results and proceed with super-resolution generation or switch to alternative methods, thus preventing quality deterioration when registration fails
Solution Approach 2:
The patent implements a feedback mechanism where the reliability of geometric transformation parameters is continuously evaluated based on the consistency between predicted and actual low-resolution images. This feedback loop allows the system to adaptively adjust the weighting between predicted and actual images, ensuring high-quality output even when registration accuracy varies
2Speed
If geometric transformation parameters are computed from low-resolution images to predict super-resolution images, then the processing speed is improved, but the accuracy of the predicted image deteriorates
Solution Approach 1:
The patent changes the parameter representation by working in the low-resolution domain for geometric transformation computation rather than directly in the high-resolution domain. This parameter transformation enables faster computation while maintaining prediction accuracy through subsequent reliability evaluation and weighted combination with actual low-resolution images
3Manufacturing precision
If multiple low-resolution images are used to generate a high-resolution image, then the image quality is improved, but the complexity of the processing system increases
Solution Approach 1:
The patent extracts and evaluates the reliability of geometric transformation parameters separately from the main super-resolution generation process. By isolating this evaluation step, the system can make informed decisions about which images to use and how to weight them, simplifying the overall processing logic while maintaining high image quality
Solution Approach 2:
The patent applies partial action by selectively using only the reliable portion of geometric transformation parameters for super-resolution generation. When reliability is low, the system relies more on actual low-resolution images rather than predicted images, thus avoiding the complexity of handling unreliable transformations while still achieving quality improvement from multiple images
Data Source
AI summary
A geometric transformation parameter computing unit computes a geometric transformation parameter which represents the geometric transformation of an image from the previous frame to the current frame, and calculates the level of reliability in that geometric transformation parameter. A super-resolution image prediction unit generates a prediction of a super-resolution image in the current frame by transforming a super-resolution image in the previous frame on the basis of the geometric transformation parameter. A super-resolution image generation unit calculates a low-resolution image in the current frame from the prediction result of the super-resolution image in the current frame by a simulation, calculates a difference between a low-resolution image, which is an input image for the current frame, and the simulation result, and calculates a weighted mean between a result of upsampling the difference and the prediction result of the super-resolution image in the current frame. Herein, the weighted mean is calculated on the basis of the level of reliability in the geometric transformation parameter.