Registration of dental radiographs using projective geometry
Related Experiment Video
Updated: Nov 9, 2026

Reliability of Artificial Intelligence-Based Cone Beam Computed Tomography Integration with Digital Dental Images
Published on: February 23, 2024
J Ostuni1, E Fisher, P van der Stelt
1Department of Biomedical Engineering, Rutgers University, Piscataway, New Jersey, USA.
Abstract:
Dunn and van der Stelt recently introduced a new model of radiographic image registration which makes use of the correspondence of 3D structures (Dunn SM, van der Stelt P. Dentomaxillofac Radiol 1992; 21: 142-7; Dunn SM et al. Dentomaxillofac Radiol 1993; 22: 77-80). Using a synthetic projective geometry, an algorithm was developed that utilized five-point projective invariants to describe the relative location of each image pixel. A limitation of this synthetic approach is that there will always be a single line passing through the image that cannot be registered. In this paper, an algorithm which is based upon an analytical projective geometry is introduced. This algorithm describes each image pixel by using a coordinate system called a reference triangle. It was shown experimentally that this algorithm successfully registers the entire image. It can also perform the registration almost six times faster than the original algorithm.

