Intra Block Copy Buffer Resetting for Valid Reference Vectors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video coding technologies face challenges in efficiently managing buffers and block vectors for intra block copy mode, leading to complex processing, invalid block vectors, inefficient coding, and unclear handling of CTU sizes smaller than 128x128, particularly in screen content coding.
Innovation Solution
Implementing a regular buffer for intra block copy mode that stores reference samples before and after loop filtering, using motion information from reconstructed blocks within the same video region without referencing a reference picture, and applying block vectors to determine valid reference locations within the buffer.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If a buffer is used to store reference samples for intra block copy mode, then coding efficiency is improved, but buffer management complexity increases
Solution Approach 1:
The buffer is divided into multiple regions including a current CTU buffer region and a reference CTU buffer region. This segmentation allows independent management of different buffer areas, reducing overall buffer management complexity while maintaining coding efficiency through targeted reference sample storage.
Solution Approach 2:
The buffer is initialized with valid reference samples before intra block copy mode is used. This preliminary action ensures that when block vectors are generated, valid reference samples are already available, reducing the need for complex validation and error handling during decoding.
2Loss of information
If block vectors are used for intra block copy mode, then motion information is captured, but invalid block vectors occur
Solution Approach 1:
The system checks whether reference samples at the location indicated by the block vector are valid before using them. This feedback mechanism ensures that only valid block vectors are used for prediction, preventing errors while maintaining motion information capture capability.
Solution Approach 2:
The buffer is pre-filled with valid reference samples from previously decoded CTUs before intra block copy mode is applied. This preliminary preparation ensures that block vectors always point to valid data, eliminating invalid block vector issues while preserving motion information.
3Adaptability or versatility
If CTU sizes smaller than 128x128 are used, then video coding flexibility is improved, but buffer handling becomes unclear
Solution Approach 1:
The buffer structure is designed to work with any CTU size, not just 128x128. The same buffer management principles apply regardless of CTU dimensions, providing universal handling that simplifies implementation while maintaining coding flexibility for different video resolutions and formats.
Solution Approach 2:
The buffer is organized with specific regions allocated for current and reference CTUs, allowing efficient handling of smaller CTU sizes by maintaining appropriate local buffer allocations. This local quality approach ensures proper buffer management adapts to different CTU dimensions without requiring complete redesign.
4Measurement precision
If reference samples are stored in buffer, then prediction accuracy is improved, but memory usage increases
Solution Approach 1:
The buffer is segmented into current CTU and reference CTU regions, allowing memory to be allocated only where needed for prediction. This segmentation reduces overall memory usage while maintaining prediction accuracy by storing reference samples only in the necessary buffer regions.
Solution Approach 2:
The buffer manages reference samples by replacing old reference data with new reference data as decoding progresses. This discarding and recovering approach ensures that memory is used efficiently, maintaining prediction accuracy with current reference samples while freeing memory for new CTUs.
Data Source
AI summary
A method of visual media processing includes determining a size of a buffer to store reference samples for prediction in an intra block copy mode; and performing a conversion between a current video block of visual media data and a bitstream of the current video block, using the reference samples stored in the buffer, wherein the conversion is performed in the intra block copy mode which is based on motion information related to a reconstructed block located in same video region with the current video block without referring to a reference picture.