Dynamic Video Non-Uniformity Correction via Polynomial Fitting
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional video data correction techniques fail to effectively address non-uniformity in live video composition, particularly when subjects are shot against solid colored backgrounds, leading to issues like seams, brim, wrinkles, and noise, especially with camera movement and lighting variations.
Innovation Solution
A method and system that extract feature data from video input, generate correction factors using polynomial fitting models, and apply multiplicative or additive corrections to improve uniformity, immune to noise and camera movement, by accumulating weighted data horizontally and vertically to create accurate fitting models for chromakey processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional video data correction techniques are used, then processing speed is maintained, but non-uniformity correction precision deteriorates due to seams, brim, wrinkles, and noise
Solution Approach 1:
The patent applies preliminary action by pre-calculating and storing correction factors in lookup tables before actual video processing. The system pre-processes reference images to generate correction data that compensates for non-uniformities, so that during live video composition, only simple table lookups and multiplications are needed, achieving high precision correction without real-time complex computation
Solution Approach 2:
The patent replaces complex mechanical correction processes with mathematical models and polynomial fitting algorithms. Instead of using traditional hardware-based correction mechanisms that produce artifacts like seams and brim, the system uses software-based polynomial fitting to model non-uniformity patterns and generate correction factors, eliminating the harmful artifacts while maintaining processing efficiency
2Measurement precision
If complex correction algorithms are applied, then non-uniformity correction precision improves, but processing time increases
Solution Approach 1:
The system performs complex polynomial fitting and correction factor calculation in advance during system initialization or calibration phase. These pre-computed correction factors are stored in lookup tables, allowing the live video processing to use simple table lookups and multiplications, thus achieving high correction accuracy without real-time computational overhead
Solution Approach 2:
The patent implements a hybrid approach where the correction system adapts its complexity based on requirements. For standard live video composition, simple lookup table methods are used for real-time performance. When higher precision is needed, the system can dynamically switch to polynomial fitting algorithms, balancing processing time and correction accuracy according to actual needs
3Adaptability or versatility
If static correction methods are used, then device complexity is reduced, but adaptability to camera movement and lighting changes deteriorates
Solution Approach 1:
The patent incorporates feedback mechanisms where the system continuously monitors the video input for changes in lighting conditions or camera position. When deviations from the reference pattern are detected, the system automatically updates the correction factors by comparing current frame statistics with the reference model, and adjusts the polynomial fitting parameters accordingly, maintaining correction accuracy under dynamic conditions
Solution Approach 2:
The correction system is designed to handle multiple scenarios using a unified polynomial fitting framework. The same mathematical model and correction approach work for both static and dynamic conditions, as well as for different types of non-uniformity (illumination gradients, lens shading, sensor variations). This universal approach reduces overall system complexity while maintaining high adaptability across various operating conditions
Data Source
AI summary
Methods and apparatus for dynamic correction of data for non-uniformity are disclosed. Feature data are extracted from input video data that include a subject shot against a backing area in a solid color. The feature data may describe characteristics of non-uniformity in input video data. A curve is generated based on the extracted feature data, and correction factors are formed based on the generated curve. At least one of the input video data and alpha data associated with the input video data is corrected based on the correction factors.