GPM Motion Vector Selection for Inter Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing GPM motion information candidate list generated through parity check has irrationality, leading to unsuitable motion vectors for inter prediction, resulting in a loss of predictive performance.
Innovation Solution
An inter prediction method that determines a prediction mode parameter for a current block, obtains a merge motion information candidate list, constructs a GPM motion information candidate list by sequencing motion vectors, and selects first and second motion vectors from this list to determine the inter prediction value of the current block.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If motion vectors are selected from merge motion information candidate list using parity check method, then the GPM motion information candidate list can be constructed, but the reliability of motion vectors for inter prediction deteriorates
Solution Approach 1:
The patent extracts only the necessary motion vector components from the merge motion information candidate list for GPM construction, separating them from other motion information. This extraction process selects motion vectors specifically suited for GPM partition prediction while discarding irrelevant components, thereby improving both the construction efficiency and reliability of the GPM motion information candidate list.
Solution Approach 2:
The patent applies different selection criteria to different parts of the motion information candidate list. Specifically, it uses parity check method for certain positions while employing alternative selection methods for other positions where parity check may not be optimal. This local differentiation ensures that motion vectors are selected based on their specific characteristics and suitability for inter prediction, thereby improving overall reliability.
2Device complexity
If conventional motion information candidate list is used for GPM, then the construction process is simple, but the inter prediction performance deteriorates due to loss of predictive performance
Solution Approach 1:
The patent segments the motion information candidate list construction into multiple stages: first constructing the merge motion information candidate list using conventional methods, then selectively processing specific elements for GPM suitability. This segmentation allows maintaining the simplicity of conventional construction while adding targeted improvements for inter prediction performance through selective motion vector extraction and validation.
Solution Approach 2:
The patent performs preliminary filtering and validation of motion vectors before they are used in GPM construction. By pre-selecting and verifying motion vectors that are suitable for inter prediction, the system ensures high predictive performance is achieved without significantly increasing the overall construction complexity, as the filtering is done in advance using efficient criteria.
Data Source
AI summary
An inter prediction method comprising: determining a prediction mode parameter of a current block which indicates using a GPM to determine an inter prediction value of the current block, obtaining a merge motion information candidate list, wherein the merge motion information candidate list comprises one or more motion candidate information item and the motion candidate information item comprises one or two motion vector; generating a GPM motion information candidate list by using the motion vectors included in the motion candidate information items according to a sequence of the motion candidate information items in the merge motion information candidate list comprises a motion vector; selecting, from the GPM motion information candidate list, a first motion vector of a first partition and a second motion vector of a second partition in the current block; and determining the inter prediction value by using the first motion vector and the second motion vector.