Video Stream Change Detection for Low-Latency Rate Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding methods struggle to handle abrupt scene changes in video streams transmitted over limited-rate and low-latency communication links, leading to increased data rates, buffer saturation, and latency, without compromising visual quality.
Innovation Solution
A method using Mojette transforms to estimate shifts and statistical deviations between frames, allowing detection of significant changes, and adapting coding parameters such as quantization, intra-frame prediction, or starting new GOPs to mitigate rate increases.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If predictive coding is used to compress video streams, then data transmission rate is reduced, but latency increases due to buffer saturation during abrupt scene changes
Solution Approach 1:
The patent applies preliminary action by detecting abrupt scene changes between consecutive frames before coding occurs. The detection mechanism analyzes pixel differences and identifies scene changes in advance, allowing the system to prepare appropriate coding strategies (such as switching to intra-frame coding or adjusting quantization parameters) before the actual coding process, thereby preventing buffer saturation and latency increases
Solution Approach 2:
The patent implements dynamics by dynamically adjusting coding parameters based on real-time scene change detection. The system continuously monitors frame differences and adapts coding strategies on-the-fly, switching between inter-frame and intra-frame coding modes, and adjusting quantization parameters dynamically to maintain optimal performance during both stable and abrupt scene transitions
2Productivity
If quantization parameters are adjusted to maintain target rate, then compression efficiency is improved, but video quality deteriorates during abrupt scene changes
Solution Approach 1:
The patent applies parameter changes by dynamically modifying quantization parameters based on detected scene changes. When an abrupt scene change is detected, the system adjusts quantization parameters to allocate more bits to the current frame, preventing quality deterioration. This is achieved by changing the quantization step size or QBias values in response to scene change detection, thereby maintaining video quality during critical transitions while preserving compression efficiency during stable scenes
3Quantity of substance
If inter-frame prediction is used to exploit temporal redundancy, then compression ratio is improved, but performance deteriorates during abrupt scene changes
Solution Approach 1:
The patent implements feedback by using the output of scene change detection to control the selection of coding modes. The detection mechanism provides feedback about frame similarity and abrupt changes, which then feeds back into the coding decision process. This closed-loop approach allows the system to switch from inter-frame prediction to intra-frame prediction when scene changes are detected, maintaining coding performance reliability while preserving compression ratios during normal operation
Data Source
AI summary
A method for analysing the change between two frames of an uncompressed video stream, including estimating a shift and statistical deviation between all or part of the frames in at least one direction, and using the one or more computed shifts and statistical deviations to detect a significant change between the frames. A method for coding an uncompressed video stream, wherein information relating to a significant change is used to modify the uncompressed video stream or coding parameters of the uncompressed video stream, in order to avoid peaks in rate. A device implementing either of the methods, to a complete transmitting system, to a computer program product comprising program instructions allowing the steps of the methods to be executed, and to a recording medium comprising the computer program product.