AMVP Candidate List Construction with Non-Adjacent Motion Vectors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding techniques face challenges in achieving diverse motion information in the advanced motion vector predictor (AMVP) candidate list, leading to reduced coding efficiency due to the inclusion of similar motion information from spatial, temporal, and history-based candidates.
Innovation Solution
The proposed solution involves constructing an AMVP candidate list that includes non-adjacent candidates derived from units or blocks not adjacent to the current prediction unit (PU). These non-adjacent candidates are added to the list after temporal candidates and/or before history-based candidates, potentially increasing the list length beyond two candidates.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If only adjacent spatial, temporal, and history-based candidates are included in the AMVP candidate list, then the list construction is simple and fast, but the motion information diversity is insufficient leading to reduced coding efficiency
Solution Approach 1:
The candidate list construction is segmented into multiple stages: first constructing a base list from adjacent spatial, temporal, and history-based candidates, then separately identifying and appending non-adjacent candidates from alternative candidate units. This segmentation allows the system to maintain the simplicity of traditional candidate selection while adding diversity through a structured extension process.
Solution Approach 2:
The patent extends the candidate selection from the traditional spatial-temporal-history dimension by introducing a new dimension: non-adjacent spatial candidates from alternative candidate units. This dimensional expansion allows the AMVP candidate list to incorporate motion information from blocks that are not immediately adjacent to the current prediction unit, thereby increasing motion information diversity without completely redesigning the candidate selection framework.
2Adaptability or versatility
If the AMVP candidate list length is increased beyond two candidates by including non-adjacent candidates, then motion information diversity is enhanced, but the signaling overhead and processing complexity increase
Solution Approach 1:
The patent extracts non-adjacent candidates from alternative candidate units and separates them from the traditional adjacent candidate selection process. By taking out these additional candidates and appending them to the existing AMVP candidate list, the system enhances motion information diversity while maintaining a clear distinction between mandatory adjacent candidates and optional non-adjacent candidates, thereby managing signaling overhead more effectively.
Solution Approach 2:
The candidate list construction is made dynamic by allowing the list length to adapt based on availability. The base list from adjacent candidates is constructed first, then non-adjacent candidates are conditionally appended based on their availability in alternative candidate units. This dynamic approach allows the system to increase candidate diversity when beneficial while avoiding unnecessary signaling overhead when additional candidates are not available or needed.
Data Source
AI summary
An example device includes memory configured to store the video data and one or more processors implemented in circuitry and communicatively coupled to the memory. The one or more processors are configured to determine at least one of a temporal candidate or a history-based candidate and determine at least one non-adjacent candidate, wherein the at least one non-adjacent candidate is from a unit that is not adjacent to a current prediction unit (PU). The one or more processors are configured to determine an advanced motion vector predictor (AMVP) candidate list including the at least one of the temporal candidate or the history-based candidate and the at least one non-adjacent candidate. The at least one non-adjacent candidate is added to the AMVP candidate list after the temporal candidate or before the history-based candidate. The one or more processors are configured to code the current PU based on the AMVP candidate list.