Intra Prediction Subblock Splitting for Video Encoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video encoding methods using spatial intra prediction in H.264/AVC experience decreased compression efficiency due to inaccurate prediction of pixels farther from adjacent pixels, leading to increased differences between actual and predicted pixel values.
Innovation Solution
The method involves predicting a current block by splitting it into subblocks and using pre-reconstructed neighboring subblocks for intra prediction, transforming the residual subblocks into the spatial domain, and adding the predicted subblocks to the residual subblocks to reconstruct the current block, thereby improving prediction accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If spatial intra prediction is used to predict current block from previously encoded adjacent pixels, then encoding efficiency is improved, but prediction accuracy decreases for pixels farther from adjacent pixels
Solution Approach 1:
The current block is divided into multiple sub-blocks, and prediction is performed separately for each sub-block using locally available adjacent pixels. This segmentation allows each sub-block to use nearby pixels for prediction, improving accuracy while maintaining encoding efficiency.
Solution Approach 2:
Different prediction approaches are applied to different regions (sub-blocks) within the current block. Each sub-block uses its local adjacent pixels for prediction, ensuring that the prediction quality is optimized for each local region rather than using a uniform approach for the entire block.
2Loss of energy
If adjacent pixels are used for predicting current block, then compression efficiency is improved, but difference between actual and predicted pixel values increases
Solution Approach 1:
The block is segmented into sub-blocks to reduce the prediction distance. By predicting each sub-block separately using its local adjacent pixels, the difference between actual and predicted values is reduced, improving pixel value accuracy while maintaining compression efficiency.
Solution Approach 2:
Adjacent pixels are reconstructed and stored before being used for prediction of the current block. This preliminary reconstruction ensures that the most accurate pixel values are available for prediction, reducing the error between actual and predicted values.
3Device complexity
If previous blocks are used for prediction, then encoding complexity is reduced, but prediction reliability decreases for distant pixels
Solution Approach 1:
The current block is divided into sub-blocks that are predicted using locally adjacent pixels from previous blocks. This segmentation maintains low encoding complexity by using simple adjacent pixel copying while improving prediction reliability for each sub-block by reducing the distance to reference pixels.
Data Source
AI summary
A method performed by an apparatus for intra-predicting a current block in the unit of subblocks, comprises: determining an intra prediction mode of the current block among a plurality of intra prediction modes; identifying the subblocks split from the current block; and sequentially reconstructing the identified subblocks using an intra prediction mode identical to the determined intra prediction mode of the current block, wherein a subblock to be reconstructed is predicted using at least one pre-reconstructed neighboring subblock in the current block.