Scene-Change Detection via Prediction Error Entropy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current scene-change detection methods in video signals are inadequate for accurately identifying changes between frames coded at different dynamic ranges, particularly in high-definition and ultra-high-definition video streams, which can lead to suboptimal rendering on various display devices.
Innovation Solution
The method involves constructing scene-change detectors using prediction error characteristics and entropy measures between video signals coded at different dynamic ranges, employing prediction models and entropy calculations to identify sudden changes in distortion and variance, thereby enhancing detection accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional scene-change detection methods are used on video signals coded at different dynamic ranges, then the detection process is simple, but the detection accuracy deteriorates due to inability to accurately identify changes between frames with different color grades
Solution Approach 1:
The detection method is segmented into multiple independent components: prediction error calculation between corresponding frames, entropy computation on the error signals, and scene-change decision logic. This segmentation allows each component to be optimized independently while maintaining overall accuracy for detecting scene changes across different dynamic ranges.
Solution Approach 2:
Prediction error signals serve as an intermediary between the two video signals coded at different dynamic ranges. By computing the difference between predicted and actual frames, the method creates a standardized intermediate representation that eliminates dynamic range differences, enabling accurate scene-change detection through entropy analysis of these error signals.
2Adaptability or versatility
If video signals are coded at different dynamic ranges to support both legacy and advanced display devices, then compatibility is improved, but scene-change detection accuracy deteriorates due to color grade differences
Solution Approach 1:
Instead of attempting to detect scene changes directly in the heterogeneous video signals with different dynamic ranges, the method inverts the approach by predicting one frame from the other and analyzing the prediction errors. This inversion transforms the problem from detecting changes in incompatible signals to analyzing the entropy of standardized error signals, thereby maintaining both compatibility and detection accuracy.
Solution Approach 2:
The method changes the parameter space by transforming the detection problem from the original video signal domain to the prediction error domain. By operating on derived error signals rather than the original heterogeneous video data, the system achieves parameter standardization that enables accurate scene-change detection across different dynamic ranges while preserving support for both legacy and advanced display devices.
Data Source
AI summary
A scene change is determined using a first and a second video signal, each representing the same scene or content, but at a different color grade (such as dynamic range). A set of prediction coefficients is generated to generate prediction signals approximating the first signal based on the second signal and a prediction model. A set of prediction error signals is generated based on the prediction signals and the first signal. Then, a scene change is detected based on the characteristics of the prediction error signals. Alternatively, a set of entropy values of the difference signals between the first and second video signals are computed, and a scene change is detected based on the characteristics of the entropy values.