Video reconstructions in dentistry

To present two practical techniques for three-dimensional (3D) modeling of the human jaw from a sequence of intra-oral images.

A data acquisition system consists of: 3D digitizing arm, CCD camera and a laser projector in addition to a software module of two 3D modeling techniques; shape from shading (SFS) and space carving (SC).

Several experiments have been conducted on a sample of students at the Computer Vision and Image Processing (CVIP) Laboratory at the University of Louisville, Louisville, KY. Other experiments were performed on solid models of human jaw.

The SFS technique, using perspective projection and camera calibration, extracts the 3D information from a sequence of two-dimensional images of the jaw. Data fusion of range data and 3D registration techniques develop the complete jaw model. The SC approach is implemented on a sequence of calibrated images. On the two reconstructions, we fit a mesh model to the data, in order to create a solid 3D model.

The accuracy of the reconstructed 3D model of human jaw is calculated based on the measurements on real jaws.

The SFS-based technique seems to provide more faithful information about the shape of the tooth tops. However, the SC algorithm successfully reconstructed 3D models of the human jaw with sub-millimeter accuracy, which is as accurate as (or even better than) the first technique without using any range measurements or laser projectors. The average error in distance calculation was found to be 0.74 mm, which is an acceptable resolution for many orthodontics and maxillofacial applications.

Accurate 3D reconstruction of the human jaw enables many orthodontics and dental imaging research findings to be applied directly to a digital jaw model--not to a cast--using computer vision and medical imaging tools.