Video Quality Indicators From Gradients and Lateral Averages
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for calculating video quality information are inadequate in terms of accuracy and efficiency, particularly in assessing the subjective quality of video signals transmitted via telecommunication networks.
Innovation Solution
A device and method that utilize non-linear transformations based on absolute gradient values and lateral average values of video frames to calculate video quality indicators, incorporating formulas like R = (max(0, (E-S))/(E+S+c(E)) and c(E) = 0.5 * (c0 + mean(E)) to enhance accuracy and efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If existing methods for calculating video quality information are used, then the calculation process is simple, but the accuracy and efficiency of video quality assessment is insufficient
Solution Approach 1:
The patent segments the video quality assessment process into distinct computational stages: gradient calculation, lateral average computation, non-linear transformation application, and final quality indicator derivation. This segmentation allows each component to be optimized independently while maintaining overall accuracy and efficiency.
Solution Approach 2:
The patent transforms the video quality assessment by changing parameters from direct pixel comparisons to gradient-based features, then to lateral average values, and finally applying non-linear transformations. This parameter transformation sequence significantly improves measurement precision by capturing perceptually relevant features.
2Productivity
If existing methods for calculating video quality information are used, then the computational resources required are low, but the efficiency of video quality assessment is insufficient
Solution Approach 1:
The patent extracts only the most relevant features for quality assessment: absolute gradient values and lateral average values. By taking out and focusing on these specific features rather than processing all video data, the method achieves high efficiency with reduced computational resource consumption.
Solution Approach 2:
The patent performs preliminary computations of gradient values and lateral averages before applying the final non-linear transformation. This preliminary action organizes and pre-processes the data in an efficient manner, reducing the computational burden during the actual quality assessment phase.
3Measurement precision
If simple calculation methods are used, then the computational load is low, but the alignment with human perception is insufficient
Solution Approach 1:
The patent applies non-linear transformations to the gradient and lateral average values to better align with human perceptual characteristics. This parameter change from linear to non-linear domain captures the complex ways humans perceive video quality, including sensitivity to different types of distortions.
Solution Approach 2:
The patent computes quality indicators locally at each pixel position using gradient and lateral average values, then aggregates these local measurements. This local quality approach allows the assessment to capture spatially varying quality characteristics that correlate with human perception of different regions in the video.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A device (10) for calculating video quality information of a video sequence is provided. The device (10) comprises a data acquisition unit (11) for acquiring at least one video frame with respect to the video sequence, and a calculation unit (12) connected to the data acquisition unit (11). In this context, the calculation unit (12) is configured to calculate at least one video quality indicator value on the basis of a non-linear transformation with the aid of at least one absolute gradient value with respect to the at least one video frame and at least one lateral average value with respect to the at least one video frame.