Image Decoding With Adaptive Merge and Inter-Prediction Modes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The increasing demand for high-resolution, high-quality images and videos, particularly in virtual and augmented reality, requires more efficient compression techniques to reduce transmission and storage costs.
Innovation Solution
Implementing a method and apparatus that derive prediction samples using a default merge mode when a merge mode is not selected, applying a regular merge mode, and utilizing inter prediction modes such as MMVD, CIIP, and partitioning modes to enhance image coding efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If high resolution and high quality image/video are transmitted or stored, then image quality is improved, but transmission and storage costs are increased
Solution Approach 1:
The patent uses prediction samples (copies) derived from reference blocks to represent the current block, avoiding transmission of actual pixel data. The decoder reconstructs the current block by combining prediction samples with residual information, effectively copying visual information from already-transmitted reference data.
Solution Approach 2:
The patent changes the representation from raw pixel values to parameter-based prediction (motion vectors, merge indices, residual coefficients). By encoding motion parameters and residuals instead of full pixel data, the patent achieves high-quality reconstruction with reduced data volume, directly addressing the cost-quality tradeoff.
2Productivity
If multiple advanced prediction modes (MMVD, merge subblock, CIIP, partitioning) are implemented, then compression efficiency is improved, but device complexity is increased
Solution Approach 1:
The patent divides the current block into multiple prediction modes and processes different regions with different prediction techniques. For example, partitioning mode splits the block into two regions with different motion characteristics, allowing specialized prediction for each region while maintaining overall compression efficiency.
Solution Approach 2:
The patent implements a dynamic prediction mode selection mechanism where the encoder chooses from multiple prediction modes (merge, MMVD, merge subblock, CIIP, partitioning) based on content characteristics. The decoder dynamically adapts to the selected mode using mode indication information in the bitstream, achieving high compression efficiency without requiring all modes to be permanently active.
3Measurement precision
If a default merge mode is applied when merge mode is not selected, then prediction accuracy is improved, but encoding complexity is increased
Solution Approach 1:
The patent pre-generates a merge candidate list containing multiple potential reference blocks before the actual prediction process. When merge mode is not selected, the decoder can immediately use the default merge mode with pre-computed candidates, avoiding the need for complex real-time candidate generation and improving prediction accuracy while managing encoding complexity through advance preparation.
Data Source
AI summary
A decoding apparatus includes at least one processor configured to acquire image information including inter prediction mode information and residual information through a bitstream; generate residual samples based on the residual information; generate prediction samples of a current block by applying a prediction mode determined based on the inter prediction mode information; and generate reconstructed samples based on the prediction samples and the residual samples. The inter prediction mode information includes at least one of a first flag related to whether a subblock merge mode is applied, a second flag related to whether a regular merge mode is applied, a third flag related to whether a merge mode with motion vector difference (MMVD) is applied or a fourth flag related to whether a combined inter-picture merge and intra-picture prediction (CIIP) mode is applied.