Geometric Partition Weighting for Faster Motion Vector Derivation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding technologies face inefficiencies in deriving motion vectors for geometric partitioning modes, particularly in handling complex partition shapes and reducing computational complexity in merge candidate selection and redundancy checks.
Innovation Solution
A method for video encoding and decoding that involves partitioning a coding block into parts and determining a weight index at a sample position to derive a blending weight for blending prediction signals, using a motion index from a motion storage unit, and storing the motion field based on the determined motion vector type, without relying on lookup tables.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If motion vectors are derived using existing video coding technologies for geometric partitioning modes, then coding performance can be achieved, but computational complexity increases and efficiency decreases
Solution Approach 1:
The coding block is divided into multiple geometric partitions (e.g., triangles, wedges, or other shapes) based on partition parameters. Each partition is then processed independently to derive motion vectors, which reduces the computational complexity compared to processing the entire block as a single unit while maintaining coding performance.
Solution Approach 2:
Instead of deriving motion vectors for all possible partitions exhaustively, the method selects a subset of partitions based on partition parameters and derives motion vectors only for those partitions. This partial action approach reduces computational complexity while maintaining sufficient coding performance.
2Productivity
If merge candidate selection and redundancy checks are performed with existing methods, then coding accuracy is maintained, but processing time increases
Solution Approach 1:
The method performs preliminary actions by pre-calculating partition parameters and determining geometric partitions before motion vector derivation. This allows for more efficient merge candidate selection and redundancy checks during the main processing loop, reducing overall processing time while maintaining coding accuracy.
Solution Approach 2:
The method implements optimized redundancy checks that can skip unnecessary computations. By using partition parameters to determine which redundancy checks are necessary, the method can rush through or skip redundant verification steps, thereby reducing processing time without compromising coding accuracy.
3Adaptability or versatility
If complex partition shapes are handled with conventional methods, then geometric flexibility is achieved, but device complexity increases
Solution Approach 1:
The method uses partition parameters (such as partition angle, partition distance, or other geometric parameters) to define complex partition shapes. By changing these parameters, the system can adapt to different geometric configurations without increasing implementation complexity, as the same parameter-based framework handles all partition shapes uniformly.
Data Source
AI summary
In a method of video decoding at a video decoder, a bitstream of a coded video associated with a current coding block that is partitioned into a first part and a second part along a partition edge for prediction in a geometric partitioning mode is received. A weight index at a sample position of the current coding block is determined based on a look up table. The weight index is used to derive a blending weight for blending two prediction signals corresponding to the first part and the second part of the current coding block. A stored motion vector type for a first motion storage unit in a motion field of the current coding block is determined based on a motion index of the first motion storage unit. A motion vector of the first motion storage unit is stored based on the stored motion vector type.