Video Codec Evaluation With Polynomial Curve Fitting
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing codec performance evaluation tools require monotonically increasing input data and overlapping intervals, limiting their versatility and accuracy in evaluating codec performance, especially when data sets do not conform to these conditions.
Innovation Solution
A method involving polynomial fitting with constraints to handle non-monotonic data and non-overlapping intervals, using a cubic polynomial to fit anchor and test data points, allowing for accurate calculation of BD-PSNR and BD-rate across various bitrates and quality metrics.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If existing codec performance evaluation tools are used, then evaluation can be performed under standard conditions, but the tools cannot handle non-monotonic data and non-overlapping intervals, limiting versatility and accuracy
Solution Approach 1:
The patent transforms the evaluation approach by changing the mathematical parameters from requiring monotonic sequences to using polynomial fitting that can handle non-monotonic data. The key parameter change is replacing the assumption of monotonic increasing data with a polynomial model that can capture complex relationships between bitrate and quality metrics, thereby improving both adaptability to different data conditions and measurement precision through more accurate curve fitting
Solution Approach 2:
The patent creates a universal evaluation method that works across multiple scenarios: monotonic and non-monotonic data, overlapping and non-overlapping intervals, and various bitrate ranges. By using polynomial fitting as a general framework, the tool achieves multi-functionality in handling diverse codec performance evaluation conditions that were previously addressed by separate specialized tools
2Measurement precision
If polynomial fitting with constraints is used, then accurate evaluation can be achieved with non-monotonic and non-overlapping data, but the computational complexity increases
Solution Approach 1:
The patent applies partial polynomial fitting by focusing on fitting curves only within relevant bitrate intervals rather than attempting to fit the entire dataset globally. This partial action approach reduces computational complexity while maintaining accuracy in the regions that matter most for codec performance evaluation, avoiding the excessive computation that would result from global fitting across all data points
Solution Approach 2:
The patent segments the bitrate range into multiple intervals and applies polynomial fitting separately to each segment. This segmentation strategy breaks down the complex global fitting problem into simpler local fitting tasks, reducing overall computational complexity while improving accuracy by capturing local characteristics of the bitrate-quality relationship in each segment
Data Source
AI summary
Methods, apparatus, and computer readable storage medium evaluating codec performance. One method includes obtaining m anchor data points each generated based on a respective anchor encoded video bitstream; obtaining n test data points each generated based on a respective encoded test video bitstream, n being an integer; fitting the m anchor data points with an anchor curve, the anchor curve being based on an anchor polynomial, wherein the anchor polynomial is monotonic in an x-axis range; fitting the n test data points with a test curve, the anchor curve being based on a test polynomial, wherein the test polynomial is monotonic in the x-axis range; and evaluating the test codec performance based on the anchor curve and the test curve, to obtain an evaluation result.