Markerless 3D Motion Capture via Cross-Sectional Image Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing motion capture systems are costly, cumbersome, and often fail to operate in real-time due to the need for markers or sensors on subjects and the analysis of data from multiple cameras, limiting their deployment and use.
Innovation Solution
A method and system that captures motion and determines the shape and position of objects in 3D space using cross-sectional images obtained from reflections or shadows, without the need for sensors or markers, by slicing images into 2D cross-sections and reconstructing the 3D structure based on light sources and intersection points.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If markers or sensors are worn by the subject, then motion capture accuracy is improved, but device complexity and cost increase
Solution Approach 1:
The patent extracts and removes the markers and sensors from the motion capture system, achieving motion capture without these traditional components. The system uses natural body features and computer vision algorithms to track motion, eliminating the need for wearable devices while maintaining measurement capability.
Solution Approach 2:
The patent creates a digital model (copy) of the subject's body with key anatomical landmarks identified through image processing. This virtual model is then used to track motion by comparing successive images, replacing the need for physical markers or sensors on the actual subject.
2Measurement precision
If multiple cameras are used to capture images from different angles, then measurement precision is improved, but device complexity and cost increase
Solution Approach 1:
The patent segments the motion capture task into distinct computational steps: image acquisition, feature detection, landmark identification, and 3D reconstruction. This allows the system to achieve accurate spatial relationship capture through software processing rather than requiring complex multi-camera hardware configurations.
Solution Approach 2:
The patent replaces the mechanical/optical system of multiple cameras with a computational approach using computer vision algorithms. A single camera or minimal camera setup combined with advanced image processing substitutes for the traditional multi-camera mechanical system, reducing hardware complexity while maintaining measurement precision.
3Measurement precision
If data from multiple cameras is analyzed and correlated, then motion capture accuracy is improved, but processing time increases
Solution Approach 1:
The patent performs preliminary action by pre-identifying and tracking anatomical landmarks in each image frame before performing the full 3D reconstruction. Key body points are detected and tracked across frames in advance, which accelerates the subsequent motion capture processing while maintaining accuracy.
Solution Approach 2:
The patent implements real-time processing by skipping intermediate computational steps and using optimized algorithms that directly compute 3D positions from 2D image data. The system rushes through the data processing pipeline using efficient mathematical formulations that avoid iterative optimization, enabling real-time motion capture without sacrificing measurement precision.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
Enables real-time, cost-effective motion capture without the need for sensors or markers, improving the efficiency and practicality of capturing complex movements in various applications.
Implementation Method 1
cross-sectional images obtained from reflections or shadows
Implementation Method 2
cross-sectional images obtained from reflections or shadows
Data Source
AI summary
Methods and systems for determining a gesture command from analysis of differences in positions of fit closed curves fit to observed edges of a control object to track motion of the control object while making a gesture in a 3D space include repeatedly obtaining captured images of a control object moving in 3D space and calculating observed edges of the control object from the captured images. Closed curves are fit to the observed edges of the control object, including control object appendages for multiple portions of any complex control objects, as captured in the captured images by selecting a closed curve from a family of similar closed curves that fit the observed edges of the control object as captured using an assumed parameter. Using fitted closed curves, a complex control object is constructed from multiple portions of any complex control objects and one or more of control object appendages appended.