Motion Vector Predictor Derivation for Static-Area Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing moving picture coding methods using inter prediction lack efficiency in deriving motion vector predictors, particularly for static areas, leading to suboptimal coding performance.
Innovation Solution
A method that derives both first and second motion vector predictor candidates, where the second candidates have a predetermined vector, such as a zero vector, to enhance the selection process, and includes an index in the bitstream to identify the chosen predictor, allowing for improved coding efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If conventional motion detection is used to derive motion vectors, then the coding process can be completed with basic inter prediction, but coding efficiency is insufficient particularly for static areas
Solution Approach 1:
The motion vector predictor candidates are segmented into two distinct groups: first candidates derived from spatially or temporally adjacent blocks, and second candidates with predetermined vectors (such as zero vectors). This segmentation allows the coding system to handle different area types (moving vs. static) with appropriate predictor types, improving overall coding efficiency without overwhelming complexity
Solution Approach 2:
The patent pre-defines second motion vector predictor candidates with predetermined vectors (e.g., zero vectors) in advance, before the actual coding process. This preliminary preparation ensures that static areas have suitable predictors readily available, eliminating the need for complex real-time derivation and improving coding speed and efficiency
2Reliability
If only spatial or temporal adjacent blocks are used for predictor derivation, then the derivation process is simple, but the number of viable predictor candidates is limited
Solution Approach 1:
The patent merges two different sources of motion vector predictor candidates into a unified candidate set: candidates from adjacent blocks (spatial/temporal) and candidates with predetermined vectors. This combination increases the diversity and number of viable predictors, improving error resistance by providing more options for accurate prediction in various coding scenarios
Solution Approach 2:
The patent introduces a new parameter dimension for motion vector predictors by including predetermined vectors (such as zero vectors) that differ fundamentally from adjacency-based predictors. This parameter change expands the solution space and provides more versatile prediction options, particularly beneficial for static areas where zero motion is common
Data Source
AI summary
A moving picture coding apparatus for coding a current block from among blocks in a picture is provided. A motion vector predictor candidate list, having a fixed size, is generated, with the motion vector predictor candidate list including first and second candidates. The motion vector predictor candidate list is generated by: deriving the first candidate from a first motion vector used to code a first block, with the first block being adjacent to the current block; and deriving the second candidate that has a second motion vector that is a non-zero value vector based on an X-Y axis of a fixed offset value, with the second motion vector not being derived by coding a block adjacent to the current block and the fixed offset value being a non-zero value that is added in a picture header and commonly used for the blocks in the picture.