Video Quality Assessment Without Reference Signals
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current automated image evaluation systems for video quality rely on comparisons to a reference pattern, which is impractical and subjective, and do not effectively identify degradation without a golden standard signal, especially in multiplexed video transmission where quality can degrade at various points.
Innovation Solution
An automated method and system that evaluates video impairments like noise by analyzing differences between nearby pixels using a processor-equipped imaging device, such as a set top box, which applies filters like the 2nd order Laplace filter to determine the ratio of noisy pixels and compares it to pre-determined thresholds, allowing for the assessment of video quality without a reference image.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If automated image evaluation systems compare transmitted signals to a template or golden standard signal, then measurement precision is improved, but device complexity and loss of information worsen because the original video or template is generally not available at an arbitrary place for evaluation
Solution Approach 1:
The system performs self-service by using the video signal itself to evaluate its own quality. The method analyzes the video content to detect noise and degradation without requiring an external reference image, allowing the system to autonomously assess transmission quality at any point in the distribution network
Solution Approach 2:
The evaluation process is segmented into analyzing individual pixel blocks and comparing them with adjacent blocks. This segmentation allows quality assessment to be performed on local regions independently, enabling reference-free evaluation while maintaining measurement precision through localized statistical analysis
2Measurement precision
If subjective testing is used to evaluate video quality by human viewers, then measurement precision is improved, but productivity and loss of time worsen because subjective testing is time consuming and not useful for operational monitoring
Solution Approach 1:
The patent replaces the mechanical system of human subjective testing with an automated electronic analysis system. The method uses digital signal processing techniques, specifically Laplace filtering and statistical analysis of pixel blocks, to objectively measure video quality parameters, thereby eliminating time consumption while maintaining assessment accuracy
Solution Approach 2:
The system changes the measurement parameters from subjective human perception to objective mathematical metrics. By calculating the standard deviation of Laplace-filtered pixel blocks and comparing against threshold values, the system transforms quality assessment into quantifiable parameter analysis that can be performed rapidly and automatically
3Ease of operation
If PSNR metrics are used to measure video quality, then ease of operation is improved, but measurement precision worsens because PSNR does not always rank quality of an image or video sequence in the same way that a person would
Solution Approach 1:
The patent applies local quality assessment by dividing the video image into blocks and analyzing each block independently. The method calculates quality metrics for local regions and compares them with adjacent blocks, capturing local degradation patterns that global PSNR metrics miss, thereby improving quality ranking accuracy while maintaining automated operation
Solution Approach 2:
Instead of comparing the test signal to a reference signal as PSNR does, the patent inverts the approach by comparing adjacent blocks within the same test signal. This inversion allows the system to detect degradation through internal inconsistencies in the video content itself, achieving better correlation with human perception without requiring reference material
Data Source
AI summary
A method for measuring the amount of noise in a video image includes receiving a signal from an imaging device; extracting a luma component from a color image; applying a filter to compute the second derivatives of the extracted luma component; determining a ratio of noise pixels to total pixels in the second derivative; and comparing the ratio to a pre-determined ratio.