Shared Motion Estimation Cost Metrics for Overlapping Video Units
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional motion estimation systems face high complexity and inefficiency in exploring all combinations of coding unit/prediction unit configurations for tree-coded video compression, leading to increased processing power and area requirements.
Innovation Solution
The system employs shared motion estimation cost metrics for overlapping units, processing different unit sizes in parallel pipelines, sharing initial stages for cost metric calculations, and refining motion vectors in both shared and parallel stages to achieve efficient motion estimation at multiple levels of granularity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If all combinations of coding unit/prediction units are explored to find the best quality video coding, then video coding quality is improved, but coding complexity increases by a factor of N*M
Solution Approach 1:
The patent divides the motion estimation process into multiple hierarchical levels, where different unit sizes are processed in parallel pipelines. Each pipeline handles specific coding unit sizes independently, segmenting the overall complex problem into manageable sub-problems that can be solved simultaneously, thereby reducing the exponential complexity to a more manageable level.
Solution Approach 2:
The patent merges the motion estimation processes for different unit sizes by sharing initial stages among parallel pipelines. The cost metric calculation stages are combined and shared across all pipelines, allowing the system to process multiple unit sizes efficiently while reusing computational resources, thus reducing overall complexity.
2Productivity
If multiple pipelines process different unit sizes in parallel, then processing efficiency is improved, but hardware area increases
Solution Approach 1:
The patent combines multiple motion estimation pipelines into a shared architecture where initial cost metric calculation stages are common to all pipelines. This merging approach allows parallel processing of different unit sizes while significantly reducing the total hardware area required, as the shared stages serve multiple purposes simultaneously.
Solution Approach 2:
The shared motion estimation cost metric stages are designed to be universal and multi-functional, serving all parallel pipelines for different unit sizes. This multi-functionality allows the same hardware blocks to be reused across multiple processing contexts, reducing overall hardware area while maintaining parallel processing efficiency.
3Area of stationary object
If shared motion estimation cost metrics are used for overlapping units, then hardware area is reduced, but motion estimation precision may be compromised
Solution Approach 1:
The patent segments the motion estimation process into distinct hierarchical levels, with shared cost metric calculation stages followed by separate refinement stages for each unit size. This segmentation ensures that while initial cost metrics are shared to save area, the final precision-critical refinement steps are performed separately for each unit, maintaining motion estimation accuracy.
Solution Approach 2:
The patent implements a dynamic refinement process where motion vectors are initially estimated using shared cost metrics, then separately refined for each unit size based on specific requirements. This dynamic approach allows the system to adapt the level of refinement for each unit type, ensuring precision is maintained where needed while benefiting from area savings through sharing.
Data Source
AI summary
An apparatus having a processor and a circuit is disclosed. The processor may be configured to (i) compare, at a first level of a motion estimation hierarchy, first units of a current picture with a reference picture to generate first metrics, (ii) combine, at the first level, the first metrics to generate second metrics and (iii) refine, at a second level of the hierarchy, the first metrics and the second metrics to generate motion vectors. Multiple metrics may be refined in parallel. The first metrics generally correspond to the first units in an overlapping unit of the current picture. The second metrics generally correspond to a plurality of second units in the overlapping unit. Each second unit may overlap one or more first units. The circuit may be configured to process the overlapping unit based on the motion vectors to generate an output signal.