Spatial Filtering of Motion Vectors for Video Encoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional video encoders are inefficient due to high computational complexity, particularly in producing high-quality video, as they perform many brute force computations to determine motion vectors, leading to slow processing.
Innovation Solution
The method employs spatial filtering and hierarchical motion search, leveraging massively parallel computing to make adjacent motion vectors similar, reducing bit-rate by setting differentials to zero, and using GPUs for parallel processing to improve video quality and compression efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional video encoders perform brute force computations to determine motion vectors, then video quality is improved, but processing speed deteriorates
Solution Approach 1:
The patent applies preliminary action by performing a coarse motion search first to obtain initial motion vectors, then using these vectors as starting points for refinement. This preliminary coarse search narrows the search space for subsequent detailed optimization, avoiding exhaustive brute force computation while maintaining high video quality through iterative refinement of motion vectors.
2Measurement precision
If conventional video encoders perform exhaustive motion estimation, then motion vector accuracy is improved, but computational complexity increases
Solution Approach 1:
The patent segments the motion estimation process into multiple stages: a coarse search phase that divides the search space into manageable regions, followed by refinement phases that focus computational effort on promising candidates. This segmentation allows the system to achieve high motion vector accuracy without requiring exhaustive search of all possible motion vectors, thereby reducing computational complexity.
Solution Approach 2:
The patent applies partial action by performing motion estimation on selected candidate blocks rather than all blocks, and by limiting the refinement process to a fixed number of iterations or until convergence criteria are met. This approach achieves sufficient motion vector accuracy for high quality video without the excessive computational cost of exhaustive search.
3Adaptability or versatility
If conventional video encoders use traditional motion estimation methods, then compatibility with video standards is maintained, but encoding efficiency deteriorates
Solution Approach 1:
The patent applies dynamics by implementing an adaptive motion estimation process that dynamically adjusts search parameters, refinement iterations, and candidate selection based on local image characteristics and complexity metrics. This dynamic approach maintains compatibility with standard video codecs while significantly improving encoding efficiency by concentrating computational resources on difficult regions and using simpler methods for easy regions.
Data Source
AI summary
Embodiments include a video data encoding method comprising receiving video input frames, and performing motion estimation on the video received frames. The motion estimation comprises performing a hierarchical motion search to find motion vectors with optimum sum of absolute difference (SAD) values, and performing spatial filtering of the motion vectors, wherein spatial filtering comprises making some pairs of motion vectors the same to achieve a zero differential.