AI Image Coding With Asymmetric Scaling for Low-Bitrate Reconstruction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image encoding and decoding technologies struggle to efficiently handle high-resolution/high-quality images, leading to increased bitrate requirements and potential information omission during compression and reconstruction.
Innovation Solution
An AI-based encoding and decoding method using deep neural networks (DNNs) for down-scaling and up-scaling images, allowing for different resolution ratios in horizontal and vertical directions, thereby reducing bitrate and preventing information omission.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If conventional image encoding is used for high-resolution images, then image quality is maintained, but bitrate increases significantly
Solution Approach 1:
The image is divided into two resolution components: a low-resolution base image and a high-resolution detail image. The base image is encoded at low resolution to reduce bitrate, while the detail image captures only the essential high-frequency information. This segmentation allows the system to achieve high image quality at low bitrate by combining both components.
Solution Approach 2:
The patent extracts and encodes only the most important high-frequency detail information from the high-resolution image, rather than encoding the entire high-resolution image. This extraction approach retains the essential visual quality while dramatically reducing the amount of data that needs to be transmitted.
2Ease of manufacture
If conventional image encoding is used, then encoding process is simple, but information omission occurs during compression and reconstruction
Solution Approach 1:
The patent employs feedback mechanisms where the encoder and decoder exchange side information about the image characteristics and compression parameters. This feedback loop allows the system to adapt the encoding process to preserve critical information, preventing information omission while maintaining encoding feasibility.
Solution Approach 2:
The system dynamically adjusts encoding parameters such as quantization levels, transform coefficients, and detail image resolution based on the original image characteristics. This parameter adaptation ensures that information is preserved effectively during compression, preventing omission while managing encoding complexity.
3Quantity of substance
If AI-based down-scaling is applied to reduce resolution, then bitrate is reduced, but image resolution decreases
Solution Approach 1:
The patent applies AI-based down-scaling as a preliminary step before encoding. The deep neural network performs intelligent down-scaling to create a low-resolution base image that retains essential information. This preliminary action prepares the image for efficient encoding while preserving the capability to reconstruct high resolution through the detail image component.
Solution Approach 2:
The system transitions from a single-resolution image to a multi-resolution representation by adding a detail image dimension. The base image provides the low-resolution framework, while the detail image adds high-frequency information in an additional dimensional space, enabling reconstruction of high resolution without increasing bitrate.
Data Source
AI summary
An AI decoding apparatus includes a memory storing instructions and a processor configured to execute the instructions to obtain AI data related to AI down-scaling of an original image and image data generated as a result of encoding a first image, obtain a second image corresponding to the first image by decoding the image data, determine a resolution ratio in a horizontal direction and a resolution ratio in a vertical direction between the original image and the first image, based on the AI data, and obtain, by an up-scaling deep neural network (DNN), a third image in which a resolution in at least one of a horizontal direction and a vertical direction is increased from the second image based on the resolution ratio in the horizontal direction and the resolution ratio in the vertical direction.