Spatial Motion Vector Scaling Using Temporal Predictor
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Scalable video coding techniques face increased processing costs due to the need for division operations in spatial motion vector scaling and inter-layer position mapping, especially when dealing with layers of different spatial scalability in High Efficiency Video Coding (HEVC) and its extensions.
Innovation Solution
The implementation of a method that uses a temporal scaling function, such as the Temporal Motion Vector Predictor (TMVP) scaling function, to perform spatial motion vector scaling without division operations, and introduces scaling parameters for horizontal and vertical directions to optimize scalability ratios, reducing computational and memory requirements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If spatial motion vector scaling is performed using conventional methods, then motion vector accuracy is improved, but processing complexity increases due to division operations
Solution Approach 1:
The patent changes the parameter representation by using scaled integer arithmetic instead of floating-point division. Motion vectors are represented with scale factors (e.g., 1/16, 1/32, 1/64) and integer operations are used to perform scaling through bit shifting and multiplication, avoiding division operations while maintaining precision.
Solution Approach 2:
The patent replaces the mechanical division operation with an equivalent computational mechanism using multiplication and bit shifting. The division-free approach substitutes the traditional division-based scaling mechanism with a more efficient arithmetic mechanism that achieves the same scaling effect without the computational overhead of division.
2Measurement precision
If division operations are used for spatial scaling, then scaling precision is improved, but computational cost increases
Solution Approach 1:
The patent transforms the scaling operation from division-based to multiplication-based with bit shifting. By representing scale factors as powers of two (1/16, 1/32, 1/64), the patent enables precise scaling through efficient bit manipulation operations that maintain precision while reducing computational power requirements.
Solution Approach 2:
The patent substitutes the computationally expensive division mechanism with a more efficient multiplication and bit-shifting mechanism. This replacement maintains scaling precision while significantly reducing the computational power needed to perform the operation.
3Measurement precision
If arbitrary division operations are performed for inter-layer position mapping, then mapping accuracy is improved, but processing time increases
Solution Approach 1:
The patent changes the arithmetic parameters by using integer-based scaling with predetermined scale factors instead of arbitrary division. This allows position mapping to be performed using efficient integer arithmetic operations that maintain mapping accuracy while reducing processing time through avoidance of division operations.
Solution Approach 2:
The patent performs preliminary preparation by establishing predetermined scale factors and integer representations before the actual mapping operation. This preliminary action enables the mapping to be executed using pre-computed parameters and efficient integer arithmetic, reducing the time required during actual processing while maintaining accuracy.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
In one implementation, an apparatus is provided for encoding or decoding video information. The apparatus comprises a memory unit configured to store video information associated with a base layer and/or an enhancement layer. The apparatus further comprises a processor operationally coupled to the memory unit. In one embodiment, the processor is configured to determine a scaling factor based on spatial dimension values associated with the base and enhancement layers such that the scaling factor is constrained within a predetermined range. The processor is also configured to spatially scale an element associated with the base layer or enhancement layer using the scaling factor and a temporal motion vector scaling process.