Subblock Intra Prediction with Reduced Mode Parsing Overhead
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding techniques face inefficiencies in intra prediction due to unnecessary parsing of intra-prediction modes and subblock reconstruction, particularly when blocks are subdivided, leading to increased overhead and reduced coding efficiency.
Innovation Solution
A method and device for video coding that includes decoding a subblock partition application flag, performing intra prediction based on this flag, and determining a reconstruction order for subblocks to optimize intra prediction, thereby eliminating unnecessary parsing and improving coding efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the block is subpartitioned into smaller blocks to address pixels far from reference pixels, then prediction accuracy is improved, but overhead for transmitting intra-prediction mode increases
Solution Approach 1:
The current block is divided into multiple subblocks of equal size, allowing different prediction modes to be applied to different subblocks. This segmentation enables more accurate local prediction while managing overhead through selective mode signaling.
Solution Approach 2:
Different intra-prediction modes are applied to different subblocks based on their local characteristics. Each subblock can use the most appropriate prediction mode for its specific content, improving overall prediction accuracy while maintaining controlled overhead.
2Loss of information
If a single prediction mode is transmitted for the original block before subdivision, then overhead is reduced, but prediction efficiency decreases
Solution Approach 1:
Instead of transmitting a single prediction mode for the entire block, the system applies prediction modes partially to each subblock. This allows the use of a default mode for some subblocks while using explicitly signaled modes for others, balancing overhead and efficiency.
Solution Approach 2:
The system changes the granularity of prediction mode application from block-level to subblock-level. By introducing subblock partitioning, it enables more flexible parameter (prediction mode) assignment that adapts to local variations within the block.
3Measurement precision
If reference samples are used for subblock prediction, then prediction accuracy improves, but some reference samples become useless
Solution Approach 1:
The system performs preliminary subblock partitioning and determines which reference samples will be useful before actual prediction. By pre-organizing reference samples according to subblock boundaries and prediction directions, it avoids generating useless reference samples in the first place.
Solution Approach 2:
The system extracts and uses only the relevant reference samples for each subblock's prediction, discarding or not generating useless reference samples. This selective extraction of useful reference information improves efficiency by eliminating waste.
Data Source
AI summary
A video coding method and a device using subblock partition-based intra prediction are disclosed. The video coding method and device perform intra prediction of the subblocks by eliminating unnecessary parsing of the intra-prediction mode or revising the prediction mode or reconstruction order when intra-predicting subblocks divided from the current block.