Dynamic Super-Resolution via Server-Side Database Updates
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current image super-resolution technologies face limitations in improving image quality beyond a certain amplification factor, with reconstruction-based methods requiring excessive input samples and learning-based methods being constrained by fixed and unchangeable training databases.
Innovation Solution
A dynamic super-resolution method that involves training image samples on a server, updating a local image database with a server-side database, and using template vectors and low-resolution-high-resolution dictionary blocks to enhance image quality, allowing for continuous improvement and personalized settings.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If reconstruction-based super-resolution technology is used to increase the amplification factor, then image resolution is improved, but the quantity of input image samples required increases dramatically
Solution Approach 1:
The patent applies preliminary action by pre-training a learning model using a large-scale image training set before actual super-resolution processing. This pre-training phase extracts and stores background knowledge about image structures and patterns, which can then be reused during inference without requiring additional input samples for each new image, thus resolving the contradiction between resolution improvement and sample quantity requirements
Solution Approach 2:
The patent introduces a learning model as an intermediary between input low-resolution images and output high-resolution images. This model acts as a mediator that has been trained on extensive image data to understand relationships between resolutions, allowing it to generate high-quality outputs without requiring the original input samples to be increased in quantity
2Measurement precision
If learning-based super-resolution technology is used to improve image quality, then image quality is greatly improved, but the local image database becomes fixed and unchangeable
Solution Approach 1:
The patent applies dynamics by implementing a database updating mechanism where the local image database is periodically refreshed with new images from the server. This makes the previously static database dynamic and adaptable, allowing the system to continuously improve image quality while maintaining versatility through ongoing updates with newly acquired image data
3Productivity
If a fixed local image database is used for super-resolution, then processing speed is maintained, but image quality improvement is limited
Solution Approach 1:
The patent applies preliminary action by pre-processing and organizing image data into structured databases on the server before distribution to user devices. This preliminary organization allows for efficient retrieval and updating operations, maintaining fast processing speeds while enabling continuous quality improvement through systematic database updates with new training images
Data Source
AI summary
A super-resolution method, a super-resolution system, a user equipment and a server. The super-resolution method includes: training an image sample at a server; obtaining a server-side image database; updating a local image database in a user device by using the server-side image database; and displaying an input low-resolution image as a high-resolution image.