Adaptive Quantization via Local Activity Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video encoding systems face challenges in efficiently adapting quantization parameters based on the non-uniform complexity of video frames, leading to suboptimal bit rate management and increased distortion, particularly in real-time encoding applications.
Innovation Solution
A method to determine local activity in sub-picture areas by calculating the minimum difference between pixels and their neighbors, allowing for adaptive quantization that segments the picture into non-overlapping local areas and identifies reference pixels to compute a complexity measure for each area, enabling efficient bit rate control.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If constant QP is used for encoding, then video quality remains consistent, but bit rate varies dramatically with video complexity
Solution Approach 1:
The patent applies local quality by dividing the video picture into multiple local areas and computing separate activity measures for each area. The QP is then adapted locally based on the activity measure of each region, allowing smooth areas to use higher QP (lower bit rate) while complex areas use lower QP (higher bit rate), thus stabilizing overall bit rate while maintaining quality consistency.
Solution Approach 2:
The patent implements dynamics by making QP adaptive rather than constant. The system dynamically adjusts QP values based on real-time computation of activity measures that reflect local video complexity. This dynamic adaptation allows the encoding system to respond to changing video content characteristics, balancing quality and bit rate requirements.
2Quantity of substance
If QP is increased to reduce bit rate, then bit rate drops, but fidelity decreases with increased distortion
Solution Approach 1:
The patent applies local quality by computing activity measures for different local areas and assigning different QP values accordingly. Smooth areas with low activity receive higher QP (aggressive compression, lower bit rate) while complex areas with high activity receive lower QP (preserved fidelity, higher bit rate). This localized adaptation prevents uniform quality loss across the entire image.
Solution Approach 2:
The patent implements parameter changes by dynamically adjusting QP based on the computed activity measure. The activity measure serves as a parameter that reflects local complexity, and QP is changed as a function of this measure. This parameter adaptation allows the system to optimize the trade-off between bit rate and fidelity for each local region.
3Measurement precision
If complex activity measures are used for accurate complexity estimation, then encoding precision improves, but encoding speed decreases
Solution Approach 1:
The patent applies segmentation by dividing the picture into multiple local areas and computing activity measures independently for each segment. This segmentation allows the use of relatively simple computation (minimum difference calculation) on small, manageable segments rather than attempting complex analysis on the entire image, thus maintaining speed while achieving adequate precision for each local region.
Solution Approach 2:
The patent implements partial action by using a simplified activity measure (minimum difference between pixel and neighbor) rather than a complete complexity analysis. This partial measure captures the essential complexity information needed for QP adaptation without requiring computationally expensive operations, thus maintaining encoding speed while providing sufficient accuracy for practical purposes.
Data Source
AI summary
A method and system are disclosed for determining local activity of at least a portion of a coding unit using a sum and difference between a pixel and neighboring pixels being coded. In one embodiment, the method comprises determining a minimum difference between the pixel and neighboring pixels for each pixel in the at least the portion of the coding unit and determining the local activity at least in part according to a sum of the differences between each pixel of the at least a portion of the coding unit and the neighboring pixels, wherein determining a difference between the pixel of the at least a portion of the coding unit and neighboring pixels comprises determining a minimum difference between the pixel and the neighboring pixels.