Video Encoder Partition Processing for Prediction Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image encoding technologies face challenges in efficiently processing image blocks, particularly when splitting them into non-rectangular partitions, which complicates prediction processing and increases processing amounts.
Innovation Solution
An encoder that includes circuitry and memory, which splits an image block into partitions, determines whether each partition is non-rectangular, and adjusts the prediction image generation process accordingly, either using a two-step operation for rectangular partitions or simplifying the process for non-rectangular partitions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the encoder uses a two-step operation to obtain prediction images for rectangular partitions (using first motion vector then second motion vector), then the prediction accuracy is improved, but the processing complexity increases
Solution Approach 1:
The patent applies different processing strategies to different partition types: rectangular partitions receive the full two-step motion vector processing for high accuracy, while non-rectangular partitions use a simplified single-step process. This local differentiation optimizes the balance between accuracy and complexity for each specific case.
Solution Approach 2:
Instead of applying the complex two-step process universally and then simplifying for non-rectangular cases, the patent inverts the approach by defaulting to the simplified process and only applying the complex two-step process where specifically needed (rectangular partitions), thereby reducing overall complexity while maintaining necessary accuracy.
2Productivity
If the encoder processes non-rectangular partitions using the simplified single-step operation, then the processing amount is reduced, but the prediction accuracy may be compromised
Solution Approach 1:
The patent recognizes that non-rectangular partitions have different geometric characteristics that make the simplified single-step motion vector process sufficient for their prediction needs, allowing processing efficiency to be optimized for these specific partition types without significantly compromising overall accuracy.
3Adaptability or versatility
If the encoder splits image blocks into non-rectangular partitions, then the adaptability to complex image structures is improved, but the processing complexity increases
Solution Approach 1:
The patent provides different processing paths tailored to specific partition geometries: non-rectangular partitions use the simplified single-step motion vector process, while rectangular partitions use the more accurate two-step process. This allows the system to adapt to various image structures effectively while managing processing complexity through localized optimization.
Data Source
AI summary
An encoder includes circuitry and memory. Using the memory, the circuitry, in operation, selects an encoding mode from among candidates including a decoder-side motion vector refinement (DMVR) encoding mode and a partition encoding mode. When the DMVR encoding mode is selected, the circuitry: obtains a first motion vector for a first image block; derives a second motion vector from the first motion vector using motion search; and generates a prediction image for the first image block using the second motion vector. When the partition encoding mode is selected, the circuitry: determines a plurality of partitions in a second image block; obtains a third motion vector for each partition; and generates a prediction image for the second image block using the third motion vector, without deriving a fourth motion vector from the third motion vector using motion search.