Scalable Video Coding Filter Adaptation for Complexity Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In scalable video coding, the increased memory access requirements and computational complexity due to up-sampling and motion compensation interpolation filters can hinder efficient coding, especially when the enhancement layer block is small or the filters are complex.
Innovation Solution
Implementing a processor-based system that uses simplified filters, such as 4-tap or 5-tap up-sampling and motion compensation interpolation filters, and applying Generalized Residual Prediction (GRP) techniques to reduce complexity, with options for GRP being disabled for small enhancement layer blocks to manage memory access effectively.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If up-sampling and motion compensation interpolation filters are used in scalable video coding, then video quality is improved, but memory access requirements and computational complexity increase
Solution Approach 1:
The patent changes the parameters of the filtering operations by using selective application of up-sampling and motion compensation interpolation filters based on block size thresholds. For small enhancement layer blocks, the filters are bypassed or simplified, while for larger blocks the full filtering is applied. This parameter-based selection resolves the contradiction by adapting the computational complexity to the actual quality requirements of different block sizes.
Solution Approach 2:
The patent introduces dynamic filtering where the application of up-sampling and motion compensation interpolation filters is not fixed but dynamically determined based on the size of the enhancement layer block. The system adapts the filtering intensity and memory access patterns according to the specific block being processed, thereby optimizing the balance between video quality and computational complexity for each block individually.
2Measurement precision
If complex up-sampling and motion compensation interpolation filters are applied, then prediction accuracy is improved, but memory access requirements increase
Solution Approach 1:
The patent applies local quality by differentiating the filtering treatment based on the local characteristics of enhancement layer blocks, specifically their size. Small blocks receive simplified or no filtering to reduce memory access, while large blocks receive full filtering for high prediction accuracy. This local differentiation resolves the contradiction by matching the filtering intensity to the actual prediction needs of each block.
Solution Approach 2:
The patent segments the enhancement layer blocks into different categories based on size thresholds, applying different filtering strategies to each segment. This segmentation allows the system to optimize memory access requirements for small blocks while maintaining high prediction accuracy for large blocks that benefit from complex filtering operations.
3Productivity
If Generalized Residual Prediction is applied to all enhancement layer blocks, then coding efficiency is improved, but processing complexity increases for small blocks
Solution Approach 1:
The patent applies partial action by selectively enabling Generalized Residual Prediction only for enhancement layer blocks that meet certain size criteria. For small blocks, the full GRP process is omitted or simplified, avoiding unnecessary processing complexity. This selective application maintains coding efficiency for blocks where GRP provides benefits while reducing complexity for blocks where it would be excessive.
Solution Approach 2:
The patent changes the operational parameters of Generalized Residual Prediction based on block size, transitioning from full GRP application for large blocks to simplified or disabled GRP for small blocks. This parameter-based control resolves the contradiction between coding efficiency and processing complexity by adapting the GRP intensity to the specific block characteristics.
Data Source
AI summary
An apparatus configured to code (e.g., encode or decode) video information includes a memory unit and a processor in communication with the memory unit. The memory unit is configured to store video information associated with a base layer and an enhancement layer. The processor is configured to up-sample a base layer reference block by using an up-sampling filter when the base and enhancement layers have different resolutions; perform motion compensation interpolation by filtering the up-sampled base layer reference block; determine base layer residual information based on the filtered up-sampled base layer reference block; determine weighted base layer residual information by applying a weighting factor to the base layer residual information; and determine an enhancement layer block based on the weighted base layer residual information. The processor may encode or decode the video information.


