Intra Block Copy Video Encoding with Adaptive Sub-Block Splitting
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video compression techniques are inefficient in encoding pictures with complex textures and varying edge directions, leading to increased data volume and reduced encoding efficiency.
Innovation Solution
The method involves block splitting into various shapes, including square, rectangular, and other geometric forms, using intra block copy (IBC) mode, where block vectors are determined and used to generate prediction blocks for efficient encoding and decoding.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If conventional video compression techniques are used, then encoding simplicity is maintained, but encoding efficiency deteriorates for pictures with complex textures and varying edge directions
Solution Approach 1:
The target block is divided into multiple subblocks through various splitting types (horizontal, vertical, diagonal, asymmetric) to better capture local texture characteristics and edge directions. This segmentation allows each subblock to be encoded with appropriate prediction parameters, improving overall encoding efficiency for complex textures while maintaining manageable complexity through systematic classification of splitting patterns.
2Manufacturing precision
If block splitting into various shapes is implemented, then picture quality improves for complex textures, but data volume increases due to additional syntax elements
Solution Approach 1:
Different splitting types are applied to different regions of the target block based on local texture characteristics. Homogeneous regions use simple splitting or no splitting, while complex texture regions use more sophisticated splitting patterns. This local adaptation improves picture quality where needed while minimizing the overall data volume increase from syntax elements.
Solution Approach 2:
The patent applies advanced block splitting and multiple syntax elements only when necessary for complex textures, rather than uniformly to all blocks. By selectively applying these techniques based on texture complexity assessment, the solution achieves high picture quality for challenging regions while keeping data volume increases minimal for simpler regions.
3Measurement precision
If multiple syntax elements are decoded for block splitting, then splitting accuracy improves, but decoding complexity increases
Solution Approach 1:
The patent establishes a predetermined hierarchy and dependency structure for decoding syntax elements. The first syntax element (indicating reference region splitting) is decoded first to determine the basic splitting type, followed by the second syntax element (related to specific splitting type) only when needed. This preliminary structuring of the decoding process reduces complexity by avoiding the need to decode all syntax elements in all cases while maintaining accurate splitting type determination.
Data Source
AI summary
A method for decoding a target block which has been encoded in an intra block copy (IBC) mode. The method comprises: determining a partition type of the target block by decoding at least one of a first syntax element for determining a reference area to refer to in order to partition the target block and a second syntax element related to the partition type of the target block; decoding block vector information regarding one or more sub-blocks, which have been partitioned from the target block according to the partition type, and determining a block vector corresponding to each of the sub-blocks by using the block vector information; and predicting the target block by generating and combining one or more prediction blocks from a current picture in which the target block is located by using the block vector corresponding to each of the sub-blocks.