Adaptive Bitrate Video Testing via Image Distance
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Adaptive bitrate streaming technologies face challenges in accurately estimating future channel capacity, leading to inefficient video quality and increased re-buffering events, particularly in variable network environments like mobile networks, resulting in suboptimal Quality of Experience (QoE).
Innovation Solution
A method for determining the bitrate variant of a semi-annotated video by calculating image distances between frames of different variants and a reference copy, allowing for accurate identification of the closest matching variant without overlay, enabling efficient QoE testing and measurements even without transcoders or per-variant overlays.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If adaptive bitrate streaming is used to deliver video over variable network environments, then video delivery flexibility is improved, but accuracy in estimating future channel capacity deteriorates leading to suboptimal bitrate selection
Solution Approach 1:
The system performs preliminary actions by capturing and storing reference video frames at multiple bitrate variants before actual streaming. These pre-prepared reference frames are then used for comparison during streaming to determine the current bitrate variant without requiring complex real-time channel capacity estimation, thus resolving the contradiction between delivery flexibility and measurement precision.
Solution Approach 2:
The invention creates copies of reference video frames at different bitrate variants and stores them for comparison. By copying and storing these reference frames with embedded frame numbers, the system enables accurate bitrate identification through image distance calculation without needing to estimate channel capacity in real-time, thereby maintaining both adaptability and precision.
2Measurement precision
If per-variant overlays with frame numbers are implemented, then bitrate identification accuracy is improved, but device complexity and processing requirements increase
Solution Approach 1:
The invention extracts only the essential information (frame numbers) from video frames and embeds them as simple visual overlays. This extraction approach provides sufficient information for bitrate identification without incorporating complex metadata or watermarks, thus improving bitrate identification accuracy while minimizing the increase in device complexity and processing requirements.
Solution Approach 2:
The frame number overlays serve as an intermediary element between the video content and the bitrate identification process. Rather than requiring direct analysis of complex video characteristics, the system uses these simple numerical overlays as a mediator that enables accurate bitrate determination through straightforward image distance calculation, balancing precision and complexity.
3Measurement precision
If image distance calculation is performed for all bitrate variants, then bitrate variant identification accuracy is improved, but computational time and processing resources increase
Solution Approach 1:
The system performs preliminary calculation and storage of image distances between reference frames at different bitrate variants during an offline preparation phase. By pre-computing these distance metrics and storing them alongside the reference frames, the invention eliminates the need for real-time calculation of image distances during streaming, thus maintaining high identification accuracy while minimizing computational time and processing resource usage during actual video playback.
Data Source
AI summary
A processing system may obtain a first frame of a first variant associated with a reference copy of a video, where a plurality of variants comprises copies of the video encoded at different bitrates, determine a frame number of the first frame from a visual overlay containing the first frame number, calculate a first image distance between the first frame and a frame of the reference copy having the frame number, determine, from among a plurality of image distances for frames of each variant having the frame number, a closest image distance to the first image distance, the plurality of image distances comprising image distances between frames of the variants and the frame of the reference copy having the frame number, and identify the first frame as being from the first variant in accordance with a variant associated with the first closest image distance.