Feature Restoration for Compressed Video Frame Quality Recovery
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video compression techniques struggle to effectively improve the quality of lower quality frames using high-quality frames, particularly in advanced video coding systems like H264 and H265, without significant computational overhead.
Innovation Solution
Implementing a neural network-based feature restoration model that utilizes high-quality feature data from previous frames to enhance the quality of lower quality frames, leveraging a trained feature restoration model to decode and restore image data efficiently.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of substance
If traditional video compression techniques (H264, H265) are used to compress video frames, then compression rate is improved, but image quality of lower quality frames deteriorates
Solution Approach 1:
The patent introduces a neural network as an intermediary component between the compressed low-quality frames and the final output. This neural network processes the compressed frames and reconstructs high-quality image data, effectively mediating between the compressed input and the quality output requirement
Solution Approach 2:
The system performs preliminary compression using traditional codecs (H264/H265) to achieve high compression rates, then applies a neural network in a second stage to restore quality. This preliminary compression followed by quality restoration resolves the contradiction by separating the compression function from the quality preservation function
2Manufacturing precision
If high quality frames are used to improve lower quality frames, then image quality is improved, but computational overhead increases
Solution Approach 1:
The patent changes the parameter of image quality from a fixed property to a variable that can be adjusted through neural network processing. By applying intelligent algorithms selectively to enhance quality where needed, the system improves image quality without uniformly increasing computational overhead across all frames
Solution Approach 2:
The patent replaces traditional mechanical video processing methods with neural network-based intelligent processing. This substitution enables quality improvement through learned patterns rather than computationally intensive traditional image processing techniques
Data Source
Figure 1~2
Figure 3
Figure 4
AI summary
An electronic device, including one or more processors configured to execute instructions; and a memory storing the instructions which, when executed by the one or more processors, configures the one or more processors to generate high-quality feature data of a current frame, by implementing a feature restoration model that is provided reference feature data, received by an electronic device, and corresponding to compressed feature data of a reference image corresponding to a first time that is different from a second time to which the current frame corresponds, and low-quality feature data received by the electronic device, and corresponding to compressed data of the current frame, that has a lower image quality than the reference image; and generate a current frame of a third image quality higher than the second image quality, based on the high-quality feature data.