Encoder Block Flipping and Skip Mode in Intra Block Copy Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The H.265/HEVC standard's intra block copy (BC) prediction mode is inefficient for coding blocks with predictable displacements and content with reversed patterns, and existing methods do not effectively handle these scenarios.
Innovation Solution
An encoder performs hash-based block matching to determine a reference region for a current block, accounting for flipping of the prediction region relative to the reference region, and decides on intra BC prediction in skip mode based on matching block hash values, using a hash table to identify candidate blocks with matching hash values for efficient encoding.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional intra BC prediction is used without block flipping, then the encoding process is simple, but coding efficiency is poor for content with reversed patterns
Solution Approach 1:
The patent introduces dynamic block flipping operations that adaptively transform reference blocks based on content characteristics. The encoder evaluates whether flipping operations (horizontal, vertical, or both) improve prediction accuracy for the current block, allowing the prediction process to dynamically adjust to different content patterns including reversed patterns.
Solution Approach 2:
The patent changes the orientation parameters of reference blocks by applying flipping operations. By modifying the spatial orientation (horizontal flip, vertical flip, or both) of reference blocks before comparison, the system can effectively match blocks with reversed patterns, thereby improving prediction accuracy without fundamentally changing the intra BC prediction framework.
2Measurement precision
If hash-based block matching with flipping is implemented, then prediction accuracy improves for reversed patterns, but computational complexity increases
Solution Approach 1:
The patent pre-computes hash values for candidate blocks and stores them in a hash table before the actual matching process. This preliminary organization of data allows for rapid retrieval and comparison during prediction, reducing the computational burden of evaluating multiple flipping operations for each block comparison.
Solution Approach 2:
The patent creates flipped copies of reference blocks (horizontally, vertically, or both directions) to compare against the current block. By generating these flipped versions and using hash-based matching, the system can efficiently identify the best matching reference block regardless of orientation, improving prediction accuracy for reversed patterns while managing computational complexity through the hash table structure.
3Loss of substance
If skip mode is used for intra BC prediction, then bit rate is reduced, but applicability is limited to blocks with identical matches
Solution Approach 1:
The patent extends skip mode applicability by dynamically evaluating flipped versions of candidate blocks. Instead of only comparing the current block with the original reference block, the system now also compares with horizontally flipped, vertically flipped, and both-flipped versions. This dynamic extension allows skip mode to be applied to a broader range of blocks including those with reversed patterns, thereby reducing bit rate more effectively while maintaining versatility.
4Productivity
If block flipping is considered during hash-based matching, then coding efficiency improves for predictable displacements, but processing time increases
Solution Approach 1:
The patent pre-organizes candidate blocks in a hash table with their hash values computed in advance. This preliminary structuring allows the encoder to quickly retrieve candidate blocks and evaluate flipping operations only for promising matches, rather than computing hashes and evaluating all possible blocks. The hash-based indexing significantly reduces the search space and processing time required to evaluate block flipping options.
Data Source
Figure 1~2b
Figure 3
Figure 4
AI summary
Innovations in the area of encoding blocks using hash-based block matching improve coding efficiency in many situations. For example, some of the innovations relate to encoder-side decisions about when to use block flipping during prediction. Other innovations relate to encoder-side decisions about when to use skip mode for a block predicted using intra block copy ("BC") prediction.