Object Tracking via Synthetic Image Templates
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current image recognition and tracking systems for three-dimensional objects require modification of the objects being tracked, such as emitting signals or using specific markers, and cannot effectively identify and track multiple objects using regular cameras through algorithms generated and parameterized using CAD models.
Innovation Solution
A method and system for identifying and tracking objects from two-dimensional data using a tracker-identifier that processes spatial and temporal data to provide output signals containing the object's type, position, orientation, and shape changes in three dimensions, employing spatial processing with harmonic images and k-dimensional hyperspherical harmonic weights, and utilizing a graphical processing unit for synthetic object data generation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If objects are modified with active signal emission or specific markers to enable tracking, then tracking capability is improved, but object universality deteriorates
Solution Approach 1:
The patent creates synthetic 2D images of 3D objects from CAD models and uses these synthetic images as templates for recognition. Instead of modifying physical objects with markers, the system copies the visual appearance of objects through computational generation and matching algorithms, enabling tracking of unmodified objects.
Solution Approach 2:
The patent replaces physical tracking mechanisms (markers, active emitters) with computational methods. The system substitutes mechanical modification of objects with software-based image synthesis and pattern recognition algorithms that process 2D camera images to identify and track objects without physical modification.
2Measurement precision
If multiple sensors are used for triangulation to improve position accuracy, then measurement precision is improved, but device complexity increases
Solution Approach 1:
The patent extracts the essential visual information from 2D camera images and processes it through image synthesis and pattern recognition algorithms. This approach eliminates the need for multiple sensors while achieving accurate position and orientation measurement through computational analysis of single-camera imagery.
Solution Approach 2:
The patent substitutes physical triangulation systems with computational geometry algorithms. Instead of using multiple physical sensors to geometrically triangulate positions, the system uses image processing and synthetic image matching to derive spatial information, reducing hardware complexity while maintaining measurement precision.
3Measurement precision
If objects are designed specifically for tracking purposes to improve recognition accuracy, then measurement precision is improved, but ease of manufacture deteriorates
Solution Approach 1:
The patent uses synthetic images generated from CAD models as recognition templates. These synthetic images accurately represent objects in various poses and lighting conditions, providing high recognition accuracy without requiring physical modification or specialized design of the tracked objects themselves.
Solution Approach 2:
The system pre-generates synthetic images of objects in multiple configurations and stores them as templates for recognition. This preliminary creation of reference data allows the system to accurately recognize and track objects without requiring the objects themselves to be specially designed or modified for tracking.
Data Source
AI summary
This invention describes a method for identifying and tracking an object from two-dimensional data pictorially representing said object by an object-tracking system through processing said two-dimensional data using at least one tracker-identifier belonging to the object-tracking system for providing an output signal containing: a) a type of the object, and/or b) a position or an orientation of the object in three-dimensions, and/or c) an articulation or a shape change of said object in said three dimensions.