Scaling Factor Detection for Compressed Media
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for detecting the scaling of compressed videos or images fail to accurately identify the original resolution, leading to unaccounted scaling impairments and biased Quality-of-Experience (QoE) scores, which reduces the comparability of QoE across different profiles.
Innovation Solution
A method involving frequency domain transformation, statistical computation, and scoring functions is applied to distinguish spikes caused by up-scaling from those due to compression, allowing for the identification of the original resolution by modeling features in both horizontal and vertical directions of the input video or images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If frequency domain transformation and statistical computation are applied to detect scaling, then measurement precision of original resolution is improved, but device complexity increases
Solution Approach 1:
The patent extracts specific statistical features (mean, standard deviation, skewness, kurtosis) from the frequency domain data obtained through DCT transformation. By focusing on these key statistical parameters rather than analyzing the entire frequency spectrum, the system achieves accurate scaling detection while reducing computational complexity. The extraction of distinctive statistical characteristics allows differentiation between compression artifacts and upscaling effects.
Solution Approach 2:
The patent transforms the video data from spatial domain to frequency domain using DCT, changing the parameter representation. This parameter transformation enables the detection algorithm to work with frequency coefficients instead of pixel values, making it easier to distinguish scaling artifacts from compression artifacts through statistical analysis of frequency domain characteristics.
2Measurement precision
If comprehensive statistical analysis is performed in frequency domain, then detection accuracy is improved, but processing time increases
Solution Approach 1:
The patent segments the frequency domain analysis by dividing it into horizontal and vertical direction components. By computing separate statistics for horizontal and vertical frequency coefficients, the system can efficiently analyze different orientation patterns of scaling artifacts. This segmentation allows parallel processing and reduces the computational burden compared to analyzing the entire 2D frequency spectrum simultaneously.
Solution Approach 2:
The patent computes only the essential statistical parameters (mean, standard deviation, skewness, kurtosis) needed for scaling detection rather than performing exhaustive analysis of all frequency coefficients. This partial action approach focuses computational resources on the most discriminative features, achieving sufficient detection accuracy without the time cost of complete frequency domain analysis.
3Reliability
If scaling detection is implemented to improve QoE scoring, then reliability of quality assessment is improved, but ease of operation deteriorates
Solution Approach 1:
The patent implements self-service by automatically detecting the original resolution and scaling factors without requiring manual input or configuration. The system autonomously performs frequency domain transformation, statistical computation, and scaling factor determination, then integrates this information into QoE scoring. This automation maintains high reliability while preserving ease of operation, as users simply need to provide the compressed video input without needing to understand or configure the detection algorithm.
Data Source
AI summary
Detection of scaling of compressed videos or images is provided. A frequency domain transformation is applied along both horizontal and vertical directions of input video or images to generate frequency domain data. Statistics in the frequency domain data are computed for each of the horizontal and vertical directions to extract features. The features are modeled to scores along each of the horizontal and vertical directions. An original resolution of the input video or images in the horizontal and vertical directions is identified according to the scores.