Body Tracking with Super Points and Probabilistic Fitting
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing body tracking systems using depth sensors and cameras face challenges in accurately determining the shape and position of a moving body, particularly in handling deformations and self-intersections, with existing methods relying on explicit point correspondences or deformation models that are inefficient and prone to noise.
Innovation Solution
A system that uses a depth sensor and computational device to fit data points to a body model with a probabilistic fitting algorithm, where 'super points' are identified and given additional weight, and constraints such as self-intersection, angle, and pose prior constraints are applied to improve accuracy and robustness.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If explicit point correspondences or deformation models are used to determine body shape and position, then the tracking system can handle body movements, but the system becomes inefficient and prone to noise
Solution Approach 1:
The patent extracts and removes the problematic deformation models and explicit point correspondence requirements from the tracking system. By using a point cloud-based approach with superpoints, the system eliminates the need for complex deformation modeling while maintaining tracking capability, thereby improving both reliability and efficiency
Solution Approach 2:
The patent changes the fundamental parameters of the tracking approach by transitioning from model-based parameters (deformation models, point correspondences) to data-driven parameters (point cloud density, superpoint weights). This parameter change enables the system to achieve high accuracy without the computational burden of traditional methods
2Measurement precision
If traditional fitting algorithms are used without super points, then the system is simpler to implement, but measurement precision deteriorates due to noise and deformations
Solution Approach 1:
The patent applies local quality by identifying superpoints with higher weights in regions of the point cloud that correspond to important body features. This localized enhancement of point quality improves measurement precision for critical areas without requiring complex algorithms throughout the entire system
Solution Approach 2:
The patent performs preliminary action by pre-identifying superpoints and assigning them higher weights before the actual fitting process. This preprocessing step improves measurement precision without adding complexity during the main tracking operation, as the superpoint identification is done once during initialization
3Reliability
If constraints such as self-intersection, angle, and pose prior are applied to the probabilistic fitting algorithm, then tracking robustness improves, but computational complexity increases
Solution Approach 1:
The patent applies partial action by selectively enforcing only the most critical constraints (self-intersection and angle constraints) during the fitting process, rather than applying all possible constraints. This selective approach maintains tracking robustness while avoiding the excessive computational complexity of implementing all constraint types
Solution Approach 2:
The patent substitutes mechanical constraint enforcement with a probabilistic approach where constraints are incorporated as soft constraints in the cost function. This substitution replaces rigid mechanical constraint checking with a more flexible probabilistic framework that achieves similar robustness with reduced computational overhead
4Productivity
If GPU processing is used to achieve real-time performance, then processing speed improves, but hardware requirements and system complexity increase
Solution Approach 1:
The patent uses a computationally efficient algorithm design that can be implemented on standard CPUs without requiring expensive GPU hardware. The use of superpoints and optimized fitting procedures enables real-time performance with simpler, more accessible hardware, eliminating the need for costly GPU processing units
Data Source
AI summary
Systems, methods and apparatuses for tracking at least a portion of a body by fitting data points received from a depth sensor and/or other sensors and/or “markers” as described herein to a body model. For example, in some embodiments, certain of such data points are identified as “super points,” and apportioned greater weight as compared to other points. Such super points can be obtained from objects attached to the body, including, but not limited to, active markers that provide a detectable signal, or a passive object, including, without limitation, headgear or a mask (for example for VR (virtual reality)), or a smart watch. Such super points may also be obtained from specific data points that are matched to the model, such as data points that are matched to vertices that correspond to joints in the model.