Browse Prior Art Database

Visual feedback after dataset optimization during 3D surface reconstruction

IP.com Disclosure Number: IPCOM000246954D
Publication Date: 2016-Jul-18
Document File: 4 page(s) / 126K

Publishing Venue

The IP.com Prior Art Database

Abstract

Certain Intraoral 3D scanners use structured light to observe the 3D surface of the teeth and build a 3D representation needed by the technician for CAD/CAM design of a restorative or orthodontic appliance. The handheld IO 3D scanner needs algorithms to properly align all the acquired 3D surfaces to build the 3D representation. Those algorithms will sometimes make errors (because the surface is flat, because soft tissues have moved or because similar features exist in the patient jaw). Dentists complain that they don't see any guidance during scan acquisition where 3D reconstruction problems occur. This local surface confidence correction algorithm shows or highlights to the dentist 3D reconstruction regions that contain problems during the 3D acquisition scan. The dentist can take immediate corrective actions during the acquisition scan. The highlighted region will disappear once the algorithm has corrected 3D reconstruction with added data collected by the dentist.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 44% of the total text.

Page 01 of 4


1. Title Visual feedback after dataset optimization during 3D surface reconstruction


2. Useful for Intraoral Camera/scanner

Certain Intraoral 3D scanners use structured light to observe the 3D surface of the teeth and build a 3D representation needed by the technician for CAD/CAM design of a restorative or orthodontic appliance. The handheld IO 3D scanner needs algorithms to properly align all the acquired 3D surfaces. Those algorithms will sometimes make errors (because the surface is flat, because soft tissues have moved or because similar features exist in the patient jaw).

One problem is that dentists complain that they don't see have any guidance, that they don't see where problems occur. When the dentists reach the final presentation screen, surface has cracks due to uncorrected stitching errors and they need to go back to the acquisition. This is not time-efficient and has a negative impact on the customer experience.


3. Correction

This correction algorithm shows regions that contain problems during the 3D acquisition scan. The dentist can take immediate corrective actions during the acquisition scan. The highlighted region will normally disappear once the algorithm has corrected the surface.

One exemplary correction algorithm is to estimate a local surface confidence and indicate to the dentist/user where the confidence is too low.

The picture below shows a 3D surface and a highlighted region.

Fig 1. (left) a gap between surfaces, (right) corrected surface after more data is scanned in the gap region.


Page 02 of 4

Fig 2. (left) 2 identical scan bodies stitched together, (right) views are automatically separated after the user scans the base between the two scanbodies.

The correction algorithm can have the following components:

- While a 3D handheld scanner is used to capture 3D surfaces, the 3D surfaces are stitched together to form a 3D dataset.

- A 3D representation of this 3D dataset is displayed to the dentist/user in real time (e.g., the capture rate of the scanner or the processing rate). The dentist sees typically a colored 3D representation of the object being scanned.

- Periodically (typically every second), several algorithms attempt to detect stitching errors and fix the 3D dataset. This triggers an update of the 3D representation (the displayed surface changes).

- A visual feedback to the user is displayed to show regions with detected problems in the 3D dataset. This visual feedback is modified each time after the algorithm above has fixed the dataset. This visual feedback can be updated at a slower rate for multiple reasons: (i) algorithms that fix the dataset need enough new data to take a different decision, (ii) to avoid generating lots of updates which might not be all exactly identical (fluctuations in the representation which is a discomfort), (iii) to avoid displaying 'transient' problems occurring while the dataset is being optimized (the goal is to provide a visual feedback on the optimized surface...