Intraoral Scanner 3D Data Determination Using Machine Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current digital intraoral scanners using structured light 3D imaging face challenges in capturing accurate 3D data due to the high reflectivity and translucency of teeth, which reduces contrast in the structured light pattern, necessitating a coating with opaque powder to enhance visibility.
Innovation Solution
The method involves projecting a light pattern with structured light projectors, capturing images with cameras, and using trained machine learning models to determine 3D coordinates by processing candidate points and their probabilities, allowing for accurate depth determination without the need for coatings.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If structured light 3D imaging is used to capture intraoral scans, then digital dental impressions can be obtained, but the high reflectivity and translucency of teeth reduce contrast in the structured light pattern, degrading measurement precision
Solution Approach 1:
The patent introduces a machine learning model as an intermediary processing layer between the structured light projection and 3D reconstruction. The model analyzes the captured light pattern images and probabilistically determines correspondences between projector rays and camera points, effectively mediating the degradation caused by tooth reflectivity and translucency without requiring physical modification of the teeth
Solution Approach 2:
The patent changes the processing parameters by using probability-based matching instead of direct geometric correspondence. The machine learning model outputs probability values for each candidate point-ray pairing, allowing the system to select the most likely correspondences even when the optical properties of teeth distort the light pattern, thereby maintaining measurement precision despite adverse optical conditions
2Measurement precision
If opaque powder coating is applied to teeth to enhance contrast, then measurement precision improves, but device complexity and ease of operation worsen due to additional preparation steps
Solution Approach 1:
The patent replaces the mechanical/physical approach of applying opaque powder coating with a computational approach using machine learning. Instead of modifying the physical state of the teeth surface, the system uses algorithms to interpret and correct for the optical effects, thereby improving measurement precision without adding physical complexity or operational steps
Solution Approach 2:
The machine learning model enables the scanning system to self-correct for the adverse optical properties of teeth. The system automatically analyzes the captured images, determines probable point-ray correspondences, and reconstructs accurate 3D data without requiring external intervention such as powder application, thus maintaining precision while reducing complexity
3Measurement precision
If opaque powder coating is applied to teeth, then measurement precision improves, but productivity decreases due to additional preparation time
Solution Approach 1:
The patent substitutes the time-consuming mechanical process of powder application with a computational method. The machine learning model processes the captured images and determines 3D coordinates directly from the raw structured light patterns, eliminating the need for powder coating preparation time and thereby maintaining high productivity while achieving accurate measurements
Solution Approach 2:
The machine learning model is pre-trained on data that includes various optical conditions of teeth surfaces. This preliminary training enables the model to handle reflectivity and translucency effects during the actual scanning process without requiring real-time preparation steps, thus maintaining both high precision and productivity
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 improves the accuracy of 3D data capture by effectively processing images to determine 3D coordinates, enhancing the precision of digital dental impressions without requiring additional coatings, thus overcoming the limitations of existing technologies.
Implementation Method 1
The surface of a subject's teeth may be highly reflective and somewhat translucent, which may reduce the contrast in the structured light pattern reflecting off the teeth
Implementation Method 2
determining three-dimensional (3D) coordinates for at least some of the plurality of points in the plurality of images based on the one or more outputs of the trained machine learning model
Data Source
AI summary
Embodiments relate to techniques for determining 3D data for 2D points in 2D images using machine learning. A method comprises using one or more trained machine learning models to determine correspondence between captured points of a captured light pattern in one or more images and projected points of a projected light pattern and determining depth information for at least some of the plurality of captured points based on the determined correspondence.