3D Model Scale Determination Using Similarity Transformation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Vision-based SLAM systems face challenges in determining the correct scale of reconstructed 3D models of real environments, leading to inaccuracies in camera pose estimation and virtual overlay in augmented reality applications, particularly when dealing with multiple objects.
Innovation Solution
A method to determine a similarity transformation between 3D feature coordinates, involving camera poses and image triangulation, to accurately scale 3D features and compute correct scale factors for true size representation of real objects, using image information from multiple images captured by cameras.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If vision-based SLAM systems use triangulation to create 3D geometrical models from 2D camera images, then the models can be constructed without pre-knowledge of the environment, but the scale of the reconstructed models becomes undetermined and random
Solution Approach 1:
The patent introduces a scale reference object as an intermediary element between the camera system and the environment being mapped. This object with known physical dimensions serves as a mediator that transfers scale information from the real world into the virtual 3D model, allowing the system to maintain both adaptability (no pre-knowledge needed) and measurement precision (correct scale through the reference object)
2Adaptability or versatility
If multiple geometrical models of multiple real objects are created using the same SLAM system, then the system can track multiple objects simultaneously, but switching between models causes scale changes that affect user experience
Solution Approach 1:
The patent applies a universal scale reference across all geometrical models of different real objects. By using the same scale reference object for multiple models, the system ensures that scale information is consistent and transferable across all tracked objects, allowing seamless switching between models without scale changes affecting user experience
3Ease of manufacture
If random scale values are assigned to each geometrical model, then model creation is simplified, but camera pose estimation and virtual overlay accuracy are compromised
Solution Approach 1:
The patent performs preliminary scaling of geometrical models by detecting scale reference objects in the environment before camera pose estimation is performed. This preliminary action establishes correct scale information in advance, ensuring that subsequent camera pose estimation and virtual overlay operations are performed with accurate scale data, thereby improving accuracy without complicating the model creation process
Data Source
AI summary
The invention is related to a method of determining a similarity transformation between first coordinates and second coordinates of 3D features, comprising providing a first plurality of 3D features having first coordinates in a first coordinate system which is associated with a first geometrical model of a first real object, wherein the first plurality of 3D features describes physical 3D features of the first real object, providing a second coordinate system, providing image information associated with a plurality of images captured by at least one camera, for each respective 3D feature of at least part of the first plurality of 3D features, wherein the respective 3D feature is captured by at least two of the plurality of images, determining camera poses of the at least one camera in the second coordinate system while the at least two of the plurality of images are captured, determining for the respective 3D feature a second coordinate in the second coordinate system according to the at least two of the plurality of images and the camera poses, and the method further comprising determining a similarity transformation between the first coordinates and the second coordinates of the at least part of the first plurality of 3D features, wherein the similarity transformation includes at least one translation, at least one rotation, at least one scale and/or their combinations in 3D space.