Motion Vector Candidate Extraction for Faster Video Inter-Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video encoding methods for inter-prediction in video coding suffer from inefficiencies in calculating motion vector predictors, leading to increased computational complexity and reduced encoding speed due to the use of inaccurate and redundant candidate motion vectors.
Innovation Solution
A method is introduced to construct a motion vector candidate list by selecting a subset of candidate motion vectors based on integer-valued motion vectors and neighboring block motion vectors, and using pre-calculated availability masks to reduce computational complexity and improve encoding speed.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a comprehensive list of candidate motion vectors is used for inter-prediction, then motion estimation accuracy is improved, but computational complexity increases and encoding speed decreases
Solution Approach 1:
The patent extracts only the most relevant candidate motion vectors from the full list of possible candidates. Specifically, it selects motion vectors from neighboring blocks and temporal reference blocks, discarding redundant candidates. This extraction approach maintains motion estimation accuracy by keeping the most promising candidates while reducing computational complexity by eliminating unnecessary calculations.
Solution Approach 2:
The patent segments the candidate motion vector selection process into distinct groups: spatial neighbors (above, left, above-left blocks) and temporal references (co-located blocks in reference frames). This segmentation allows the encoder to systematically evaluate different types of candidates with different selection criteria, improving accuracy through structured evaluation while managing computational complexity through organized processing.
2Measurement precision
If a comprehensive list of candidate motion vectors is used for inter-prediction, then motion estimation accuracy is improved, but encoding speed decreases
Solution Approach 1:
The patent extracts only the most relevant candidate motion vectors from the full list of possible candidates. Specifically, it selects motion vectors from neighboring blocks and temporal reference blocks, discarding redundant candidates. This extraction approach maintains motion estimation accuracy by keeping the most promising candidates while reducing computational complexity by eliminating unnecessary calculations.
Solution Approach 2:
The patent performs preliminary identification and selection of candidate motion vectors from structured sources (neighboring blocks and temporal references) before the actual motion estimation process. By pre-organizing and pre-selecting candidates based on their spatial and temporal relationships, the encoder speeds up the subsequent motion estimation process while maintaining accuracy through systematic candidate selection.
3Measurement precision
If redundant candidate motion vectors are included in the calculation, then motion estimation accuracy is improved, but computational complexity increases
Solution Approach 1:
The patent extracts only the most relevant candidate motion vectors from the full list of possible candidates. Specifically, it selects motion vectors from neighboring blocks and temporal reference blocks, discarding redundant candidates. This extraction approach maintains motion estimation accuracy by keeping the most promising candidates while reducing computational complexity by eliminating unnecessary calculations.
Solution Approach 2:
The patent changes the selection criteria parameters for candidate motion vectors by introducing specific spatial and temporal relationships. Instead of considering all possible motion vectors, it uses parameters such as neighboring block positions and temporal reference frame offsets to selectively identify candidates. This parameter-based selection reduces computational resource consumption by focusing calculations on relevant candidates with defined spatial-temporal relationships.
Data Source
AI summary
Constructing a motion vector candidate list for inter-prediction in video coding is provided. A video encoder performs inter prediction for a frame of the video to generate prediction residuals. The inter prediction is performed using motion vectors (MVs) for blocks of the frame. To determine the motion vector of a block, the encoder determines an integer-valued MV for the block based on a reference frame and selects, based on the integer-valued MV, a subset of candidate MVs from a list of MVs including MVs of neighboring blocks. The encoder identifies a MV from the subset of candidate MVs for the block and searches in a neighborhood of the identified MV for a refined MV. The encoder calculates the prediction residuals for the block based on a reference block in the reference frame pointed by the refined MV and encodes the prediction residuals into a bitstream representing the video.