Temporal Layer Bit Allocation for Video Compression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video compression technologies fail to effectively determine bit allocation for temporal base and enhancement layers, leading to suboptimal spatiotemporal experience and inefficient use of bandwidth resources.
Innovation Solution
A method to determine bit allocation for temporal base and enhancement layers by initializing allocations based on quality differences and constant factors, with adjustments made between GOP boundaries, incorporating minimum and maximum thresholds to ensure optimal quality and stability across layers.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If bits are allocated uniformly across temporal layers, then implementation is simple, but picture quality and spatiotemporal experience deteriorate
Solution Approach 1:
The patent applies local quality by allocating different bit rates to different temporal layers based on their specific quality requirements and importance. The base layer receives a first bit rate while enhancement layers receive different bit rates according to their contribution to overall picture quality, ensuring that critical layers receive sufficient bits while maintaining overall system efficiency.
Solution Approach 2:
The patent changes the bit rate parameter dynamically across different temporal layers. By adjusting the bit rate allocation parameter according to layer importance and quality requirements, the system optimizes both picture quality and bandwidth utilization, moving away from uniform allocation to differentiated parameter settings.
2Manufacturing precision
If more bits are allocated to enhancement layers, then spatiotemporal experience improves, but bandwidth consumption increases
Solution Approach 1:
The patent optimizes the bit rate parameter for enhancement layers by analyzing the actual quality improvement contribution of each layer. By adjusting the bit rate parameter to match the diminishing returns of quality improvement, the system achieves optimal spatiotemporal experience while avoiding excessive bandwidth consumption on higher layers where marginal gains are smaller.
Solution Approach 2:
The patent applies partial action by allocating bits to enhancement layers only to the extent necessary for achieving acceptable quality improvements. Rather than maximizing bits to all enhancement layers, the system allocates bits partially and selectively based on actual quality contribution, avoiding excessive bandwidth usage on layers with diminishing returns.
3Manufacturing precision
If base layer picture quality is improved, then enhancement layer prediction accuracy improves, but overall bit budget is consumed faster
Solution Approach 1:
The patent applies local quality by ensuring the base layer receives sufficient bit allocation to maintain quality levels adequate for prediction purposes. By focusing quality resources on the base layer where they provide maximum predictive benefit, the system improves enhancement layer accuracy without proportionally increasing overall bit consumption, as enhancement layers can operate with lower bit rates when base layer quality is sufficient.
Data Source
AI summary
Disclosed herein are techniques and computer readable media containing instructions arranged to determine a bit allocation for a temporal base layer and at least one temporal enhancement layer. The techniques are enhanced to yield improved reproduced quality for the case where both a temporal base layer and at least one temporal enhancement layer is being used. An exemplary method for determining a bit allocation to one base layer includes initializing a bit allocation between the base layer and the at least one enhancement layer, determining a quality difference between the base layer and the at least one enhancement layer, expressed, for example, in the Quantizer Parameter of the last picture of each layer within a GOP, and determining a new bit allocation between the base layer and the at least one enhancement layer based on the quality difference and at least one constant factor.