Adaptive Motion Vector Range Coding for Bit Allocation Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
High-resolution and high-quality image data requires efficient compression techniques to reduce transmission and storage costs, as conventional methods struggle with the increased bit allocation for motion vector differences in image coding systems.
Innovation Solution
The implementation of an Adaptive Motion Vector Range (AMVR) method that generates and transmits a Motion Vector Difference (MVD) representative value, reducing bit allocation and enhancing coding efficiency through efficient inter prediction by deriving and encoding MVD within specific ranges.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional image coding methods are used for high-resolution images, then image quality is maintained, but transmission cost and storage cost increase due to increased bit allocation
Solution Approach 1:
The patent applies parameter changes by dynamically adjusting the motion vector resolution (precision) based on the magnitude of motion vector differences. When MVD is small, higher precision (1/4 pel) is used; when MVD is large, lower precision (1 pel) is used. This adaptive parameter adjustment reduces the average bit allocation for motion vectors while maintaining adequate prediction accuracy, thereby resolving the contradiction between image quality and bit allocation quantity.
Solution Approach 2:
The patent implements dynamics through the Adaptive Motion Vector Resolution (AMVR) mechanism that dynamically selects motion vector precision based on local motion characteristics. The system transitions between different precision levels (1/4 pel, 1/2 pel, 1 pel) adaptively, allowing the coding system to optimize bit allocation according to actual motion complexity in different regions, thus reducing overall bit requirements while preserving image quality.
2Measurement precision
If motion vector precision is increased to improve inter prediction accuracy, then prediction quality improves, but bit allocation for motion vector difference increases
Solution Approach 1:
The patent changes the motion vector precision parameter adaptively based on the magnitude of motion vector difference. When the MVD magnitude is small (indicating similar motion between current and reference blocks), higher precision (1/4 pel) is applied to maintain prediction accuracy. When MVD magnitude is large, lower precision (1 pel) suffices, reducing bit allocation. This dynamic parameter adjustment resolves the contradiction between prediction accuracy and MVD bit allocation.
Solution Approach 2:
The patent applies local quality by using different motion vector precisions for different spatial regions and blocks based on their local motion characteristics. Blocks with small MVD (stationary or slowly moving regions) use higher precision, while blocks with large MVD (fast moving regions) use lower precision. This localized adaptation optimizes the balance between prediction accuracy and bit allocation for MVD across the entire image.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
An image encoding method according to the present invention, which is performed by an encoding apparatus, comprises the steps of: generating an adaptive motion vector range (AMVR) enable flag; deriving a motion vector difference (MVD) with respect to a current block; deriving a representative MVD value of an MVD range including a value of the MVD from among a plurality of MVD ranges; generating a coded MVD corresponding to the representative MVD value; and outputting the AMVR enable flag and the coded MVD through a bit stream. According to the present invention, the amount of bits allocated to an MVD can be reduced and the overall coding efficiency can thus be increased.