Image Coding Motion Vector Predictor Derivation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional image coding methods face challenges in achieving high coding efficiency due to difficulties in deriving appropriate temporal motion vector predictors, particularly when reference pictures are long-term or short-term, leading to degradation in prediction accuracy and coding efficiency.
Innovation Solution
An image coding method that determines whether reference pictures are long-term or short-term, using a first derivation scheme without scaling for long-term references and a second scheme involving scaling for short-term references, to derive motion vector predictors from co-located blocks, ensuring they are not excessively large or small.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a single motion vector derivation scheme is used for all reference pictures, then the device complexity is reduced, but the prediction accuracy degrades when reference pictures have varying temporal distances
Solution Approach 1:
The patent applies parameter changes by switching between two derivation schemes based on the temporal distance parameter. When the reference picture is a long-term reference picture, the first derivation scheme is used; when it is a short-term reference picture, the second derivation scheme is used. This dynamic parameter-based selection resolves the contradiction by adapting the derivation approach to the specific temporal characteristics of each reference picture, thereby maintaining high prediction accuracy without requiring a single overly complex universal scheme.
Solution Approach 2:
The patent segments the motion vector derivation process into two distinct schemes: a first derivation scheme for long-term reference pictures and a second derivation scheme for short-term reference pictures. This segmentation allows each scheme to be optimized for its specific use case, preventing the need for a single complex scheme that must handle all cases suboptimally. The segmentation resolves the contradiction by dividing the problem into manageable, specialized components.
2Productivity
If motion vector predictors are derived without considering temporal distance scaling, then the processing speed is improved, but the coding efficiency degrades for short-term reference pictures
Solution Approach 1:
The patent changes the processing parameter based on the reference picture type. For short-term reference pictures, temporal distance scaling is applied to maintain coding efficiency. For long-term reference pictures, scaling is omitted to save processing time. This conditional parameter change resolves the contradiction by optimizing the balance between coding efficiency and processing time according to the specific temporal characteristics of each reference picture.
3Measurement precision
If motion vector predictors are excessively large or small, then the device complexity is reduced, but the prediction accuracy degrades significantly
Solution Approach 1:
The patent uses parameter changes through temporal distance scaling to adjust the magnitude of motion vector predictors appropriately. For short-term reference pictures, scaling reduces excessively large predictors. For long-term reference pictures, no scaling prevents excessively small predictors. This dynamic parameter adjustment resolves the contradiction by keeping motion vector predictors within appropriate ranges, maintaining prediction accuracy without requiring complex clamping or correction mechanisms.
Data Source
AI summary
An image coding method includes: deriving a candidate for a motion vector predictor from a co-located motion vector; adding the candidate to a list; selecting the motion vector predictor from the list; and coding a current block and coding a current motion vector, wherein the deriving includes: deriving the candidate by a first derivation scheme in the case of determining that each of a current reference picture and a co-located reference picture is a long-term reference picture; and deriving the candidate by a second derivation scheme in the case of determining that each of the current reference picture and the co-located reference picture is a short-term reference picture.