OTT Video Quality Analysis Using Synthetic Reference VMAF
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
There is no appropriate methodology to perform reference-based Video Quality Analysis (VQA) for live or VOD internet streams due to lack of access to source streams, varying bitrates, and the absence of automated workflows.
Innovation Solution
A system that pre-calculates Video Multimethod Assessment Fusion (VMAF) scores for media and correlates them with actual playback data to enhance and accurately analyze video quality, optimizing viewer Quality of Experience (QoE) through VQA-as-a-Service (VQAaaS).
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If reference-based VQA is performed for OTT streams, then video quality measurement accuracy is improved, but access to source streams is blocked due to encoding and packaging processes
Solution Approach 1:
The patent creates a synthetic reference stream by copying and re-encoding the test stream at the original source resolution and bitrate, then uses this synthetic reference for VQA measurement. This allows reference-based quality assessment without access to the actual source stream, resolving the contradiction between measurement accuracy and source access availability.
2Adaptability or versatility
If ABR streaming with varying bitrates is used, then network adaptability is improved, but video quality evaluation becomes inaccurate due to bitrate variations
Solution Approach 1:
The patent applies different quality assessment methods to different segments of the stream based on their characteristics. For segments where the synthetic reference matches the actual reference, full reference VQA is used. For segments with bitrate variations, the system selectively uses appropriate reference streams, ensuring accurate quality evaluation while maintaining network adaptability.
3Measurement precision
If manual video quality analysis is performed, then measurement accuracy is improved, but automation and efficiency are reduced
Solution Approach 1:
The system automatically generates synthetic reference streams, performs frame alignment, calculates VMAF scores, and correlates them with playback data without manual intervention. The automated workflow maintains measurement accuracy by implementing rigorous alignment and validation procedures, while eliminating the need for manual quality analysis.
4Measurement precision
If complex alignment and correlation processes are implemented, then VQA accuracy is improved, but system complexity increases
Solution Approach 1:
The patent divides the video stream into discrete segments and performs alignment and VQA calculation on each segment independently. This segmentation approach simplifies the overall process by breaking down complex alignment tasks into manageable units, while maintaining accuracy through systematic frame-by-frame comparison and correlation of VMAF scores with playback data.
Data Source
AI summary
This disclosure provides for automated techniques to measure full reference-based QoE or VQA-as-a-Service (VQAaaS) for an Internet video stream. Generally, the approach herein involves pre-calculating VMAF scores for given media and then correlating those scores with VMAF scores computed from actual playback segments for the given media. By leveraging the pre-calculated VMAF scores and correlating them with playback data, the system provides for enhanced and accurate video quality analysis (VQA) to enable optimization of viewer Quality of Experience (QoE).