Intra Block Copy Mirror Transform for Screen Content Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing intra block copy (IBC) mode based on a translational motion model fails to effectively utilize complex motions like mirror flipping in screen content coding, leading to inefficient encoding and redundant information, which affects picture quality and compression performance.
Innovation Solution
The introduction of an IBC-mirror mode that includes horizontal and vertical mirror flipping functions within the translational model, allowing for better prediction and reduced prediction error by determining whether to perform mirror transforms on reference blocks based on mode information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If the IBC mode uses a translational motion model, then the reference block and coding block can be exactly the same, but it cannot effectively utilize complex motions like mirror flipping
Solution Approach 1:
The patent extends the static translational motion model to a dynamic model that can adapt to different motion types. By introducing mirror flipping operations and directional motion vectors, the system can dynamically adjust the reference block transformation based on the actual motion characteristics, thereby improving both adaptability and prediction accuracy.
Solution Approach 2:
The patent changes the parameters of the motion model by introducing mirror flipping flags and directional indicators. These parameter changes allow the system to represent complex motions beyond simple translation, enabling the model to adapt to mirror flipping scenarios while maintaining the basic translational framework.
2Productivity
If the IBC mode assumes all samples share the same block vector, then the encoding process is simple, but it cannot handle complex motions effectively
Solution Approach 1:
The patent segments the motion model by introducing different block vectors for different regions or directions within a coding block. This segmentation allows each region to be predicted with its own motion characteristics, improving prediction precision while maintaining encoding efficiency through selective application.
Solution Approach 2:
The patent makes the block vector assignment dynamic by allowing different samples to have different block vectors based on motion characteristics. This dynamic approach enables the system to handle complex motions effectively while maintaining reasonable encoding complexity through structured decision-making.
3Device complexity
If mirror flipping is not considered in the IBC mode, then the encoding process remains simple, but prediction errors increase in scenarios with complex motions
Solution Approach 1:
The patent applies preliminary action by performing mirror flipping operations on reference blocks before using them for prediction. By pre-processing the reference blocks with mirror flipping when needed, the system prepares the data in advance to match the motion characteristics, thereby improving prediction accuracy without significantly increasing encoding complexity.
Solution Approach 2:
The patent changes the state parameter of reference blocks by applying mirror flipping transformations. This parameter change allows the system to adapt reference blocks to match the actual motion patterns, improving prediction accuracy while maintaining encoding simplicity through conditional application based on motion detection.
Data Source
AI summary
A prediction method for a current block and an electronic device are provided, where the method includes the following. Mode information of an intra block copy-mirror mode is obtained when in a non-merge mode of intra block copy. Determine whether to perform mirror transform on a reference block according to the mode information. A prediction block of the current block is obtained by performing mirror transform on the reference block when a determination result is performing mirror transform on the reference block.