Inter Prediction Weighting for High-Resolution Image Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image compression technologies face challenges in efficiently compressing high-resolution and high-quality images, particularly in accurately performing inter prediction and managing motion information for improved compression efficiency.
Innovation Solution
The method employs a multi-reference block mode for inter prediction, using a combination of prediction modes such as merge, skip, AMVP, and IBC, with weighted sums of reference blocks to generate final predictions, and optimizes signaling through a predefined prediction mode combination set.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional single reference block inter prediction is used, then the device complexity is low, but the prediction accuracy is insufficient for high-resolution images
Solution Approach 1:
The current block is divided into multiple prediction blocks, each using different reference blocks and prediction modes. The block is segmented horizontally or vertically, allowing each segment to be predicted from different reference blocks, thereby improving overall prediction accuracy without requiring complete redesign of the prediction system
Solution Approach 2:
Multiple prediction blocks generated from different reference blocks and modes are merged through weighted summation to form the final prediction for the current block. This combining approach leverages the strengths of multiple prediction methods to achieve superior prediction accuracy compared to using a single reference block
2Measurement precision
If multiple prediction modes are combined for multi-reference block prediction, then the prediction accuracy improves, but the signaling overhead increases
Solution Approach 1:
A predefined set of prediction mode combinations is prepared in advance and stored in both encoder and decoder. This preliminary preparation eliminates the need to signal all possible mode combinations during encoding, as the decoder already has the predefined sets available for selection, thereby reducing signaling overhead while maintaining the ability to use multiple prediction modes
Solution Approach 2:
Instead of signaling complete prediction mode information for each block, the system changes the parameter representation by using compact flags to indicate selections from predefined mode combination sets. This parameter transformation reduces the number of bits required for signaling while preserving the essential prediction information
3Productivity
If conventional motion information management is used, then the encoding process is simple, but the compression efficiency is insufficient for high-resolution and high-quality images
Solution Approach 1:
The system adds a new dimension to motion information management by introducing multiple reference blocks from different temporal and spatial dimensions. Instead of relying on a single reference block, the system utilizes reference blocks from different pictures and different regions, creating a multi-dimensional prediction space that improves compression efficiency for high-resolution images
Data Source
AI summary
An image encoding/decoding method and device according to the present disclosure may: perform bidirectional prediction on the basis of a first prediction mode to thereby generate a basic prediction block of a current block; derive an additional reference block of the current block on the basis of a second prediction mode; and calculate a weighted sum of the basic prediction block and the additional reference block to thereby generate a final prediction block of the current block.