Video Processing Using Intra Block Copy with Geometric Partitioning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding technologies face challenges in improving coding efficiency, particularly in capturing complex video content features and optimizing bitstream generation.
Innovation Solution
The proposed method employs intra block copy (IBC) with geometric partitioning mode (GPM) to predict video units, allowing for geometric division of video units into sub-partitions and prediction based on multiple hypotheses, thereby enhancing coding performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If conventional video coding techniques are used, then the coding process is simple, but the coding efficiency is insufficient
Solution Approach 1:
The video unit is divided into multiple sub-partitions through geometric partitioning, allowing independent prediction for each sub-partition. This segmentation enables more precise local prediction while maintaining overall coding efficiency, resolving the contradiction between simple processing and high coding efficiency.
Solution Approach 2:
The patent introduces multiple prediction hypotheses that can be dynamically selected based on content characteristics. The system adapts between different prediction modes (IBC-GPM with various partition geometries and hypotheses) to optimize coding efficiency for different video content, balancing complexity and performance.
2Measurement precision
If video units are coded without geometric partitioning, then the coding process is fast, but the ability to capture complex video content features is limited
Solution Approach 1:
By segmenting the video unit into geometrically partitioned sub-partitions, the method captures complex video content features more accurately. Each sub-partition can be predicted independently with appropriate hypotheses, improving measurement precision while the geometric constraints maintain processing speed.
Solution Approach 2:
The patent performs preliminary geometric partitioning and hypothesis generation before actual prediction. This preliminary organization of data and prediction models enables faster processing during encoding by pre-establishing the prediction framework, reducing coding time while maintaining feature capture accuracy.
3Productivity
If standard intra prediction is used, then the bitstream generation is simple, but the compression results are suboptimal
Solution Approach 1:
The patent changes key parameters of the prediction process by introducing geometric partitioning modes and multiple hypotheses. This transforms the standard intra prediction into a more flexible IBC-GPM approach, achieving better compression efficiency while the parameterized structure keeps bitstream generation manageable.
Solution Approach 2:
The IBC-GPM method serves multiple functions: it performs geometric partitioning, generates multiple prediction hypotheses, and selects optimal predictions. This multi-functionality improves compression efficiency by handling diverse video content types within a unified framework, while the standardized process maintains bitstream generation simplicity.
Data Source
AI summary
Embodiments of the disclosure provide a solution for video processing. A method for video processing is proposed. The method includes: obtaining, for a conversion between a video unit of a video and a bitstream of the video unit, a prediction of at least one object of the video unit by using an intra block copy (IBC), wherein the video unit is coded with an IBC-geometric partition mode (GPM) where the video unit is geometrically divided into a plurality of sub-partitions or the video unit is predicated by a plurality of hypothesizes; and performing the conversion based on the prediction of the at least one object of the video unit.