Meta Model Interpolation for Video Flicker Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing techniques face challenges in handling images with inconsistent quality, particularly in videos, due to high computational complexity and limitations in capturing lossy high-frequency components, leading to performance degradation and flicker distortion.
Innovation Solution
An image processing apparatus that obtains a meta model by interpolating pre-trained reference models based on the quality of an input image, using a training dataset with images having similar content characteristics, and applies this meta model for quality processing across frames to stabilize image quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If learning-based upscaling methods are used to improve image quality, then image quality is improved when training image characteristics match input image characteristics, but image quality significantly degrades when image characteristics differ from training assumptions
Solution Approach 1:
The patent changes the parameter of image quality characteristics by obtaining a meta model based on the quality of the input image and training the meta model using a training data set corresponding to the input image. This allows the system to adapt to different image qualities dynamically rather than using a fixed training model, resolving the contradiction between achieving high image quality and maintaining adaptability to varying image characteristics.
2Manufacturing precision
If image quality processing is performed on low-quality images to improve output quality, then image quality is improved, but computational complexity increases
Solution Approach 1:
The patent applies preliminary action by obtaining and training a meta model based on the input image quality before performing the actual image quality processing. This preliminary preparation allows the subsequent processing to be more efficient and targeted, reducing the overall computational complexity while maintaining improved image quality output.
3Manufacturing precision
If image quality processing is applied to video frames to improve quality, then image quality is improved, but flicker distortion occurs due to quality variations across frames
Solution Approach 1:
The patent uses feedback by applying the trained meta model consistently across multiple video frames, where the model adapts to the input image quality and produces stable output quality. This feedback mechanism ensures temporal consistency by maintaining the same processing approach across frames, preventing flicker distortion while still improving overall image quality.
Data Source
AI summary
An image processing method including obtaining a meta model based on a quality of an input image, training the meta model by using a training data set corresponding to the input image, and obtaining a quality-processed output image from the input image based on the trained meta model.