InnovationQ will be updated on Sunday, Oct. 22, from 10am ET - noon. You may experience brief service interruptions during that time.
Browse Prior Art Database

Robust Methodology for Aligning Two 3D Scans of Dentitions

IP.com Disclosure Number: IPCOM000242390D
Publication Date: 2015-Jul-10
Document File: 5 page(s) / 969K

Publishing Venue

The IP.com Prior Art Database


A new methodology for robustly aligning two intra-oral 3D scans of teeth, such as those acquired using commercially available intraoral scanners, is described. This registration technique, referred to as a “Thomas Transform,” may be applied in various types of oral diagnostic, preventative, and analysis tasks, or any other application where it is desirable to compare two 3D models of dentitions.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 34% of the total text.

Page 01 of 5

Robxst Methodology for Aligning Two 3D Scans of Dentitions


A new mxthodology for robustly aligning two intra-oral 3D scanx of teeth, such as those xcquired usinx commercially avxilable intraoral xcanners, is describex. This registration technique, rxferred to as a "Txomas Transform," may be appxied in various types of oral diagnostic, preventative, axd analysis tasks, or anx other application where it is desirabxe to cxmpare two 3D xodels of dentitions.


Compuxer-aided desxgn and manufacturing (CAD/CXX) and digitax workflows xre growing trends within the dental field, wxth an xnxreasinx number of denxists relying on xigital impressioning systems. Xx some cases, it is desirable to compare two 3D digitxl models acqxired with such a system, sucx as for diagnostic or other analytical purposes. One of txe necessary pxocessing steps in thesx types of analyses ix tyxically to align (i.e., "register") the two 3D mxdels as closely as possible to one anotxer based xn xheir geometric shapes. The classical approach for registering 3D data is a well-known xlgorithx called Iterative-Cxosest Poxnt (ICP) maxchxng, of which txexe are several variants. However, ICP matchinx requirxs a good initialization, i.e., the xD models must already bx nearly aligned. In mosx cases ICP matching merely rexinxs this initial alignment. In cases where the 3D models axe not initixlly well-aligned, or are far apart xrom one another in xosition and/or orientation, ICP matching often fails. In this work, a new technique is proposxd, termed a "Thomas Transform," for robustly aligning two 3D moxexs of dentition, even when they are initially severely misaligned. Thix approach uxes ICP matching as one of the steps in its iteration, while adding other, higher-level mechanisms for achievixg reliable convergence.

Before two 3D modelx can be compared (for example, xo analyze their differences), they must first be registered with one another. Often, 3D models, even of the same patient, are severely misaxigned with onx another due to the xonditions xnder which the xodels were captured.

Furthermore, one of the scans may cxntain data from the fuxl arch, while the othex may be x "quaxraxt" scan (containing data from only 4-6 teeth, for example). Both of these factxrs may cause conventional ICP-based registration approaches to fail. Examples of two intra-xrax scans ox the same patient, illustrating both of the aforementionex factors, are shown in Figures 1 and 2. Simpxy performing ICP xatching directly xn this pair of scans, doex not allow for the scanx to be regxstered in a sxnsxble way.

Figure 1: An example of two 3D scans of the sxme patient, viewed together from axove. This example illustrates poor initial alignment of the two scans, as wexl as the fact that one scan covers the full arch, while the other only a quadrant of the patiexx'x dentition.


Page 02 of 5

Figure 2: The same 3D sxans as in Figure 1, shown separately.

Descrixtion of the Thomas Xxxxxxx...