Motion Vector Encoding Using Differential Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image compression techniques, such as H.264/AVC and MVComp, face limitations in compression performance due to increased indexing side information and decoder crashes when selecting motion vectors, particularly when using a limited number of predicted candidates.
Innovation Solution
A method and apparatus for video encoding and decoding that determine a current motion vector by selecting two candidate predicted motion vectors from neighboring blocks, encoding the differential vector and its index, to efficiently compress motion vectors and improve image compression efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If the number of predicted motion vector candidates increases to improve compression performance, then the quantity of indexing side information increases, but this leads to decoder crashes and increased calculation demands
Solution Approach 1:
The patent divides the motion vector prediction process into two stages: first selecting a small set of candidate predicted motion vectors (e.g., 2 candidates) from neighboring blocks, then selecting the final predicted motion vector from these candidates. This segmentation reduces the indexing side information to just 1 bit while maintaining compression performance through adaptive selection based on differential values.
Solution Approach 2:
The patent uses a limited number of candidate predicted motion vectors (partial action) rather than exhaustively searching all possible motion vectors. By selecting only 2 candidates from neighboring blocks and using a simple index (1 bit) to identify the best match, the system achieves sufficient compression performance without the complexity and errors associated with larger candidate sets.
2Device complexity
If a limited number of predicted motion vector candidates are used to reduce indexing side information, then device complexity decreases, but compression performance becomes disadvantageously limited
Solution Approach 1:
The patent changes the selection criterion from simply using the first available candidate to selecting the candidate with the minimum differential value between the current motion vector and the predicted motion vector. This parameter change (using differential comparison) allows the system to achieve high compression performance with only 2 candidates, eliminating the trade-off between complexity and performance.
3Reliability
If motion vectors of neighboring blocks are used to determine the current motion vector, then decoder crash problems are addressed, but the number of motion vector candidates remains limited
Solution Approach 1:
The patent makes the neighboring block motion vectors serve multiple functions: they provide reliable candidates for prediction (improving decoder stability) and form the basis for adaptive selection based on differential values (enabling high compression performance). The same 2 candidates from neighboring blocks are used both to ensure reliability and to achieve superior compression through intelligent selection.
Data Source
AI summary
A video encoding apparatus using an inter-prediction includes: a video encoder to determine a current motion vector of a current block, predict the current block by using the determined current motion vector, and encode a residual block representing a difference between the predicted block and the current block; and a motion vector encoder to determine a candidate predicted motion vector set based on motion information of neighboring blocks of the current block, select a predicted motion vector of the current motion vector from among the two candidate predicted motion vectors in the candidate predicted motion vector set, and encode a differential vector representing a difference between the current motion vector and the selected predicted motion vector and a predicted motion vector index indicating the selected predicted motion vector among the two candidate predicted motion vectors.