Compressed Domain Video Quality Measurement
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video quality measurement techniques require excessive resources and provide less than optimal results, especially when assessing picture quality without a reference to the original undistorted pictures.
Innovation Solution
A method and apparatus for generating a quality measure for video signals encoded using a compression algorithm with a variable quantiser step size and two-dimensional transform, utilizing quantiser step size parameters and transform coefficients to estimate peak signal-to-noise ratio (PSNR) and blockiness without decoding the signal, by extracting necessary parameters from the compressed bitstream.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional PSNR measurement methods are used, then measurement precision is improved, but device complexity and processing resources increase excessively
Solution Approach 1:
The patent extracts only the necessary parameters (quantiser step size parameters and transform coefficients) from the compressed bitstream that are sufficient for quality assessment, rather than fully decoding the video signal. This extraction approach maintains measurement precision while significantly reducing processing complexity by working directly with compressed domain data.
Solution Approach 2:
The patent introduces an intermediary quality measurement method that operates in the compressed domain using quantiser step size parameters and transform coefficients as intermediate representations. This intermediary approach avoids the need for full video decoding while still providing accurate quality assessment through mathematical relationships between these parameters and perceived video quality.
2Measurement precision
If full video decoding is performed for quality assessment, then measurement precision is improved, but processing time and resources increase
Solution Approach 1:
The patent performs quality assessment actions preliminarily by extracting and analyzing quantiser step size parameters and transform coefficients directly from the compressed bitstream before any decoding occurs. This preliminary analysis in the compressed domain provides accurate quality metrics without requiring the time-consuming process of full video decoding.
Solution Approach 2:
The patent skips the entire video decoding process by directly utilizing compressed domain parameters (quantiser step size and transform coefficients) for quality assessment. This skipping approach rushes through the quality measurement process by establishing direct mathematical relationships between compressed parameters and quality metrics, eliminating unnecessary decoding steps.
3Measurement precision
If reference-based quality measurement is used, then measurement precision is improved, but adaptability to non-intrusive monitoring decreases
Solution Approach 1:
The patent extracts quality assessment capability from the reference signal requirement, enabling non-intrusive monitoring by working solely with parameters from the compressed bitstream of the video signal under test. This extraction removes the need for original undistorted pictures while maintaining measurement precision through alternative computational approaches using quantiser step size and transform coefficients.
Solution Approach 2:
The patent enables the video signal itself to provide quality assessment information through its own compressed domain parameters (quantiser step size and transform coefficients) without requiring external reference signals. This self-service approach allows the signal to be monitored non-intrusively using information already present in its compressed representation.
Data Source
AI summary
A measure of quality of compressed video signals is obtained without reference to the original uncompressed version, but generated directly from the coded image parameters, thereby avoiding the need to decode the compressed signal. A first measure is generated from the quantizer step size and a second measure is generated as a function of the number of blocks in the picture that have only one transform coefficient. The two measures are combined. Adjustments may be made to the step-size based measure to compensate for spatial or temporal masking effects.