Video Encoder Partitioned Block Weighted Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional video compression algorithms face inefficiencies due to the use of square or rectangular macroblock shapes that do not correspond to moving object shapes, leading to reduced coding efficiency and increased computational resources from spurious high-frequencies and numerous motion vector combinations.
Innovation Solution
The method involves partitioning video data blocks into non-parallel or curved angle partitions, determining separate motion vectors for each partition, and combining predictor values using a weighting algorithm to minimize spurious high-frequencies and computational resource utilization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If square or rectangular macroblock shapes are used for video compression, then the coding process is simple and computation resources are reduced, but coding efficiency is substantially reduced because the fixed shapes do not correspond to the outline of moving objects
Solution Approach 1:
The macroblock is divided into multiple partitions with different shapes (e.g., rectangular, triangular, trapezoidal) that can better match the contours of moving objects. This segmentation allows each partition to be coded independently with motion vectors tailored to its specific shape, improving coding efficiency while maintaining manageable complexity through systematic partitioning rules.
Solution Approach 2:
Different partitions within the macroblock are assigned different shapes and motion vectors according to the local characteristics of moving objects. This local adaptation allows the coding scheme to match object contours in different regions, improving overall coding efficiency without requiring complete redesign of the entire macroblock structure.
2Productivity
If alternative macroblock shapes using oblique straight line partitions are introduced, then coding efficiency is improved by matching object shapes, but a large number of possible partitions result in large increase in utilization of computation resources
Solution Approach 1:
The macroblock is divided into a limited number of standardized partition types (e.g., two main partitions with specific geometric relationships) rather than allowing arbitrary partitions. This segmentation reduces the search space for motion vectors while still providing sufficient flexibility to match object shapes, thereby improving coding efficiency without excessively increasing computational resources.
Solution Approach 2:
Instead of varying the number and configuration of partitions extensively, the invention changes key parameters such as the angle of partition lines and the position of partition boundaries within a constrained set of options. This parameter optimization allows adaptation to different object shapes while limiting the total number of combinations that need to be evaluated, reducing computational complexity.
3Shape
If motion vector selection is performed for every pixel based on underlying partition, then object shape matching is improved, but spurious high-frequencies are introduced in predictor values that are detrimental to coding efficiency
Solution Approach 1:
Adjacent partitions with similar motion characteristics are merged into a single partition with a common motion vector. This merging reduces the number of motion vector transitions at partition boundaries, thereby minimizing the introduction of spurious high-frequencies in predictor values while still maintaining adequate shape matching through the use of multiple partition types.
Solution Approach 2:
The partition boundaries are designed with smooth transitions rather than sharp angular changes, and motion vectors are selected to create continuous predictor values across partition boundaries. This smoothing approach reduces abrupt changes in predictor values that would otherwise introduce spurious high-frequencies, thereby maintaining coding efficiency while preserving object shape matching.
Data Source
AI summary
An apparatus (e.g., a computer device) includes a video coding system. The video coding system includes an encoder. The encoder is configured to partition a block of video data into a first partition and a second partition, the first partition and the second partition being divided by a border that is at an angle to an orientation of a row or column of pixels, determine a first motion vector for the first partition and a second motion vector for the second partition, determine a first set of predictor values based on the first motion vector and a second set of predictor values based on the second motion vector, and combine the first set of predictor values and the second set of predictor values into an overall set of predictor values based on a weighting algorithm.