Video Quality Measurement Using Quantizer Step Size and Spatial Masking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for predicting video quality in decoded video sequences, especially those encoded with variable quantizer step sizes, are inefficient and often require access to original sequences, making it challenging to accurately assess quality without reference signals.
Innovation Solution
A method that generates a quality measure by combining quantizer step-size parameters with spatially sensitive masking measures, weighted based on region types such as active and background regions, within a video decoder, allowing for no-reference quality assessment.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If Mean Opinion Score (MOS) is used to measure video quality, then accuracy of quality assessment is improved, but time consumption increases
Solution Approach 1:
The patent creates an objective quality metric that copies the human visual system's perception characteristics without requiring actual human observers. By modeling how humans perceive distortion through masking effects and spatial complexity analysis, the system achieves MOS-level accuracy through automated computational analysis of the decoded video signal properties
Solution Approach 2:
The patent replaces the mechanical process of human viewing and subjective rating with an automated computational system. The objective metric uses mathematical models to analyze quantizer step sizes, spatial complexity, and masking effects, substituting human observers with algorithmic processing that operates rapidly without time consumption associated with human evaluation
2Measurement precision
If original sequence is used for quality degradation analysis, then measurement precision is improved, but ease of operation deteriorates
Solution Approach 1:
The patent extracts the essential information needed for quality assessment directly from the encoded bitstream parameters (quantizer step sizes, coding mode indicators) and the decoded signal properties (spatial complexity, masking measures). By taking out only the necessary features rather than requiring the complete original sequence, the system achieves accurate degradation measurement while operating conveniently with only the encoded and decoded signals available
Solution Approach 2:
The patent introduces an objective quality metric as an intermediary that bridges the encoded signal and perceived quality without requiring direct access to the original sequence. This intermediary metric uses the decoded signal's properties and encoding parameters to estimate quality degradation, making the assessment process convenient while maintaining measurement precision through the mediator's computational analysis
3Device complexity
If uniform quality measure is applied to entire picture, then device complexity is reduced, but measurement precision deteriorates
Solution Approach 1:
The patent applies local quality assessment by computing masking measures and spatial complexity metrics for different regions of the picture. The system identifies active regions versus background regions and applies region-specific analysis, allowing the measurement system to focus computational resources on areas that contribute most to perceived quality while maintaining overall system efficiency
Solution Approach 2:
The patent segments the picture into different region types (active regions, background regions) based on spatial complexity and masking characteristics. By dividing the picture into segments with different quality assessment requirements, the system achieves precise spatial quality measurement without uniformly complex processing across the entire image, optimizing the balance between measurement precision and device complexity
Data Source
Figure 1a~5
Figure 2
Figure 3~4
AI summary
A method of generating a measure of quality for a video signal representative of a plurality of frames, the video signal having: an original form; an encoded form in which the video signal has been encoded using a compression algorithm utilising a variable quantiser step size such that the encoded signal has a quantiser step size parameter associated therewith and utilising differential coding such that the encoded signal contains representations of the prediction residual of the signal; and a decoded form in which the encoded video signal has been at least in part reconverted to the original form, the method comprising: a) generating a first quality measure which is dependant on said quantiser step size parameter according to a predetermined relationship; b) generating a masking measure, the masking measure being dependant on the spatial complexity of at least part of the frames represented by the video signal in the decoded form according to a first predetermined relationship; and c) generating a combined measure, the combined measure being dependant upon both the first measure and the masking measure according to a predetermined relationship, wherein, the method also includes: generating an first measure which is dependant on the prediction residual of the signal according to a first predetermined relationship; and identifying one or more regions of the picture for which the first measure exceeds a threshold, wherein the masking measure is dependant on the spatial complexity of the identified region(s) according to a predetermined relationship; and further comprises: generating a background measure which is dependant on the prediction residual of the signal according to a second predetermined relationship; and identifying one or more other regions of the picture for which the background measure is below a threshold, wherein the masking measure is dependent on the spatial complexity of the identified other region(s) according to a predetermined relationship.