Motion Vector Region Setting for Parallel Video Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In image processing, the existing methods face challenges in efficiently coding and decoding motion vectors, particularly when a single coding unit is split into smaller prediction units, as this requires fixed motion vector information from adjacent units, hindering parallel processing for larger image frames like Ultra High Definition.
Innovation Solution
An image processing apparatus and method that sets spatially adjacent regions to generate spatial motion vectors in parallel, allowing for the generation and decoding of motion vectors across prediction regions within a coding unit, even when split into smaller units, by adjusting adjacent region settings based on the position of prediction regions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a single coding unit is split into smaller prediction units, then motion vector coding precision is improved, but parallel processing capability deteriorates due to dependency on fixed motion vector information from adjacent units
Solution Approach 1:
The patent divides the coding unit into multiple prediction units (e.g., 2N×N, N×2N, or N×N partitions) to improve motion vector coding precision. Each prediction unit can be processed independently with its own motion vector, allowing for more precise motion compensation while maintaining parallel processing capability through the independent region setting mechanism.
Solution Approach 2:
The patent applies different adjacent region settings based on the position of each prediction unit within the coding unit. By locally adapting the adjacent region configuration to each prediction unit's position, the patent enables independent processing of each unit while maintaining appropriate spatial relationships, thus resolving the contradiction between fine-grained partitioning and parallel processing.
2Stability of the object's composition
If motion vector information is fixed from adjacent units, then decoding consistency is improved, but processing flexibility deteriorates for large image frames
Solution Approach 1:
The patent dynamically determines adjacent regions based on the position of each prediction unit within the coding unit. Rather than using fixed adjacent regions for all prediction units, the system adapts the adjacent region selection to each unit's location, enabling processing flexibility for large image frames while maintaining decoding consistency through position-aware region setting.
Solution Approach 2:
The patent changes the parameters of adjacent regions based on the position of prediction units. By adjusting which adjacent regions are used for each prediction unit according to its position, the system maintains decoding consistency across different scenarios while adapting to various image frame sizes and prediction unit configurations.
3Device complexity
If spatial prediction motion vectors are generated sequentially, then dependency management is simplified, but processing speed deteriorates for large image frames
Solution Approach 1:
The patent segments the coding unit into multiple independent prediction units, each with its own motion vector generation process. By setting independent adjacent regions for each prediction unit, the patent enables parallel generation of spatial prediction motion vectors across multiple units, significantly improving processing speed for large image frames while keeping dependency management straightforward through the independent region configuration.
Solution Approach 2:
The patent performs preliminary setting of adjacent regions for each prediction unit based on its position within the coding unit. This preliminary configuration enables subsequent parallel processing of motion vectors without requiring complex dependency management during the actual vector generation phase, thus improving processing speed while maintaining manageable complexity.
Data Source
AI summary
The present disclosure relates to an image processing apparatus and an image processing method capable of improving process efficiency through a parallel process. In a case of corresponding PU1, there is a use of motion vector information pieces of B1, C1, and E1 which are adjacent to corresponding PU1 in the following positional relationship, and A0 and D0 which are adjacent to PU0 located on corresponding PU1 in the CU in the following positional relationship. In other words, in relation to corresponding PU1, a PU corresponding to A1 is PU0, and thus A0 is set instead of A1 as an adjacent region of corresponding PU1. The present disclosure is applicable to, for example, an image processing apparatus.