3D Scene Reconstruction Using Multi-Wavelength Texture Projection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Computer vision techniques face challenges in determining 3D scene geometry when objects have little or no visual texture, as it becomes difficult to identify corresponding features in images, hindering accurate reconstruction and manipulation tasks in robotic systems.
Innovation Solution
The method involves projecting multiple random texture patterns of light onto a scene using projectors with different wavelengths, allowing optical sensors to capture and distinguish between these patterns, enabling the identification of corresponding features and subsequent 3D reconstruction through triangulation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional computer vision techniques are used to determine 3D scene geometry, then the system can operate without additional illumination equipment, but accurate feature identification becomes difficult when objects have little or no visual texture
Solution Approach 1:
The patent introduces projected light patterns as an intermediary element between the imaging system and the textured-less objects. These projected patterns serve as artificial visual markers that enable corresponding feature identification across multiple views, thereby resolving the difficulty of detecting features on objects with little or no inherent visual texture
Solution Approach 2:
The patent employs multiple light patterns with different colors or wavelengths projected onto the scene. Optical sensors equipped with corresponding filters detect these colored patterns, creating distinctive visual signatures that facilitate accurate feature matching and 3D reconstruction even on otherwise featureless surfaces
2Measurement precision
If multiple light patterns with different wavelengths are projected onto the scene, then corresponding features can be identified more accurately, but the system complexity and equipment requirements increase
Solution Approach 1:
The patent designs the projector and optical sensor system to handle multiple wavelengths and patterns through a unified architecture. The optical sensors are equipped with filters that can selectively detect different wavelength patterns, allowing a single sensor system to process multiple light patterns without requiring separate detection systems for each wavelength
Solution Approach 2:
The system pre-projects multiple known light patterns onto the scene before capturing images. By having the patterns already established and known in advance, the system can directly compare the projected patterns against captured images to identify corresponding features, eliminating the need for complex real-time pattern generation and synchronization
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
This approach enhances the accuracy and detail of 3D reconstruction, facilitating more precise robotic manipulation tasks such as loading, unloading, and palletizing by providing reliable depth measurements and virtual representations of environments.
Implementation Method 1
projecting a plurality of different patterns of light using a plurality of projectors. The plurality of different patterns of light may include a first random texture pattern projected by a first projector of the plurality of projectors and having a first wavelength and a second random texture pattern projected by a second projector of the plurality of projectors and having a second wavelength
Implementation Method 2
receiving sensor data by a computing device and from a plurality of optical sensors. The plurality of optical sensors may be configured to distinguish between the plurality of different patterns of light
Implementation Method 3
Many computer vision techniques involve triangulating information observed from at least two known viewpoints to determine a representation of three-dimensional (3D) scene geometry. If corresponding features in two or more images of an object are identified, a set of rays generated by the corresponding points may be intersected to find the 3D position of the object or depth to the object
Data Source
AI summary
Example methods and systems for determining 3D scene geometry by projecting patterns of light onto a scene are provided. In an example method, a first projector may project a first random texture pattern having a first wavelength and a second projector may project a second random texture pattern having a second wavelength. A computing device may receive sensor data that is indicative of an environment as perceived from a first viewpoint of a first optical sensor and a second viewpoint of a second optical sensor. Based on the received sensor data, the computing device may determine corresponding features between sensor data associated with the first viewpoint and sensor data associated with the second viewpoint. And based on the determined corresponding features, the computing device may determine an output including a virtual representation of the environment that includes depth measurements indicative of distances to at least one object.