Block Vector Predictor for Intra Picture Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video coding techniques face inefficiencies in intra-picture coding, particularly in removing redundancy from text and artificial graphics regions, where repeated patterns exist, due to limitations in block vector prediction methods.
Innovation Solution
The implementation of Intra Block Copy (Intra BC) techniques, which allow for the use of a plurality of block vector predictor candidates, including neighboring blocks and default vectors, to construct a block vector candidate list and encode syntax elements indicating the selected predictor, reducing data usage by encoding differences rather than absolute block vector values.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If Intra Block Copy techniques with multiple block vector predictor candidates are implemented, then coding efficiency is improved and data required for encoding is reduced, but device complexity increases due to constructing candidate lists and processing multiple predictors
Solution Approach 1:
The block vector prediction process is segmented into multiple candidate vectors (first candidate from spatial neighboring blocks, second candidate from temporal neighboring blocks, third candidate as default). This segmentation allows the decoder to process predictions in stages, improving efficiency while managing complexity through structured organization of prediction candidates.
Solution Approach 2:
Block vector candidates are constructed and prepared in advance before the actual decoding process. The candidate list is built using previously decoded block vectors from spatial and temporal neighbors, allowing the current block to benefit from pre-computed prediction options without real-time computation overhead.
2Quantity of substance
If block vector differences are encoded instead of absolute block vector values, then data usage is reduced, but measurement precision requirements increase to accurately determine the difference values
Solution Approach 1:
Instead of encoding the complete block vector, the method extracts and encodes only the difference between the block vector and its predictor candidate. This extraction approach significantly reduces the number of bits required for encoding while the decoder can precisely reconstruct the original block vector by adding the encoded difference to the known candidate vector.
Solution Approach 2:
The encoding parameter is changed from absolute block vector values to relative difference values. This parameter transformation reduces the magnitude and variability of the encoded data, leading to more efficient compression while maintaining the ability to accurately reconstruct the original block vector through simple addition operations.
Data Source
AI summary
An example method for decoding video data includes constructing a block vector (BV) candidate list for a current block of the video data, the BV candidate list including a plurality of BV predictor candidates, each respective BV predictor candidate of the plurality of BV predictor candidates is one of: a BV of a neighboring block, a BV of a previously coded block, and a default BV; determining, based on a syntax element obtained from an encoded video bitstream, a particular BV predictor candidate from the plurality of BV predictor candidates; determining, based on the particular BV predictor candidate, a value of a BV that represents a displacement between the current block and a predictor block of video data in a picture in which the current block resides; determining, based on the value the BV, the predictor block; and reconstructing the current block based on the predictor block.