Facial Motion Capture Stabilization via Affine Transformation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current facial motion capture technologies face challenges in accurately isolating head motion from facial motion capture data, as there are no fixed landmarks on the human face to compute the overall affine transformation, leading to inconsistent results and significant errors in removing head movements during the capture process.
Innovation Solution
The method employs Horn's orientation/alignment algorithm and Random Sample and Consensus (RANSAC) to process facial motion capture data, fitting a model to the data and smoothing it to remove gross errors, thereby computing the global head transformation and isolating head motion from facial motion capture data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If hand-positioned alignment by trained artists is used, then alignment accuracy may be improved, but productivity decreases and results vary depending on artist skill
Solution Approach 1:
The patent replaces the manual mechanical alignment process performed by artists with an automated computational system using Horn's orientation/alignment algorithm. This substitutes human manual positioning with an algorithmic approach that computes the optimal affine transformation automatically, eliminating dependency on artist skill while maintaining alignment accuracy and significantly improving productivity
Solution Approach 2:
The patent transforms the alignment problem from a manual artistic task into a mathematical parameter optimization problem. By representing the alignment as an affine transformation with computable parameters (rotation, translation, scaling), the system can automatically determine optimal values using Horn's algorithm, replacing subjective artistic judgment with objective parameter calculation
2Productivity
If tracking the most stable markers is used, then head motion removal is automated, but measurement precision decreases due to significant errors
Solution Approach 1:
The patent creates a universal alignment system that works with any set of facial markers without requiring pre-selection of 'stable' markers. Horn's orientation/alignment algorithm can handle any non-collinear triplet of markers, making the system universally applicable regardless of marker stability, thus eliminating the need to identify particularly stable markers while maintaining automation
Solution Approach 2:
The patent implements an iterative refinement process where the alignment algorithm computes initial transformations, evaluates their accuracy, and refines the results. The system uses feedback from the alignment error metrics to adjust the affine transformation parameters, continuously improving measurement precision while maintaining automation
3Quantity of substance
If multiple cameras are used to capture facial motion, then motion capture completeness is improved, but device complexity and data processing difficulty increase
Solution Approach 1:
The patent segments the complex multi-camera data processing task into distinct computational stages: (1) extracting 3D point coordinates from multiple camera views, (2) applying Horn's orientation/alignment algorithm to compute the affine transformation, and (3) transforming the marker coordinates to remove head motion. This segmentation breaks down the complex processing into manageable, modular steps that can be implemented systematically
Data Source
AI summary
A method and system for removing head motion from facial motion capture data. The method includes receiving a set of measured points of a target model, wherein each point is associated with coordinates in a 3D space. The method includes computing an optimal affine transformation function. The computing includes selecting an unprocessed point from the set of measured points. The computing includes selecting two nearby neighboring points of the unprocessed point. The computing includes computing an affine transformation function that minimizes an L2-norm error. The computing includes identifying the optimal affine transformation function from a set of computed affine transformation functions. The method includes displaying an aligned target model and reference model utilizing the optimal affine transformation function. The method includes outputting the optimal affine function to a computer-readable storage medium.