AI Image Encoding and Decoding with Pre-Processing and DNN Upscaling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image encoding and decoding technologies struggle to efficiently handle high-resolution/high-quality images, particularly in terms of bitrate management and resolution adjustment, as hardware advancements demand more effective encoding and decoding methods.
Innovation Solution
Implementing artificial intelligence (AI) encoding and decoding processes that include pre-processing and the use of deep neural networks (DNNs) for up-scaling and down-scaling, with joint training of DNNs to maintain image quality and reduce bitrate through AI-based resolution adjustment.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If AI up-scaling is applied directly to decoded images without pre-processing, then image quality can be maintained, but bitrate reduction efficiency is limited
Solution Approach 1:
The patent applies pre-processing (down-scaling) to the original image before encoding, which prepares the image at an optimal resolution for compression. This preliminary action reduces the amount of data that needs to be encoded, thereby improving bitrate efficiency while maintaining quality through subsequent AI up-scaling at the decoder side
2Manufacturing precision
If resolution is increased to handle high-resolution images, then image quality is improved, but processing efficiency and bitrate increase
Solution Approach 1:
The patent segments the image processing into two distinct stages: down-scaling during encoding and AI up-scaling during decoding. This segmentation allows each stage to operate independently at optimal resolutions, improving overall processing efficiency while maintaining high output quality through the AI-based reconstruction
Solution Approach 2:
The patent changes the resolution parameter dynamically - down-scaling during encoding to reduce processing load and bitrate, then using AI up-scaling during decoding to restore high resolution. This parameter transformation approach maintains processing efficiency while achieving high-quality output
3Quantity of substance
If AI down-scaling is applied during encoding, then bitrate is reduced, but image detail information is lost
Solution Approach 1:
The patent uses AI algorithms as an intermediary during the down-scaling process to intelligently preserve important image details and features. This AI-mediated down-scaling reduces bitrate by lowering resolution while minimizing information loss through learned patterns that preserve essential visual information
Solution Approach 2:
The patent replaces traditional mechanical down-scaling methods with AI-based down-scaling that uses neural networks to intelligently process image data. This substitution allows for better preservation of image details during compression while achieving lower bitrate
Data Source
AI summary
An artificial intelligence (AI) decoding apparatus includes a memory storing one or more instructions; and a processor configured to execute the one or more instructions to obtain image data generated through first encoding on a first image; obtain a second image corresponding to the first image by performing first decoding on the image data; perform pre-processing of changing a resolution of the second image, according to a pre-determined output resolution; and obtain a fourth image having the pre-determined output resolution by applying AI up-scaling based on an up-scaling deep neural network (DNN) to a third image obtained as a result of the pre-processing.