Automatic registration of spectrophotometric retinal images

Gisli H. Halldorsson*, Jon Atli Benediktson, Gunnar Mar Zoega, Thor Eysteinsson, Einar Stefansson

*Corresponding author for this work

Research output: Contribution to journalConference articlepeer-review

4 Citations (Scopus)


This study is a part of the development of a system that enables analysis of spectrophotometric measurements of oxygen saturation in the retinal vasculature. A digital fundus camera is used to capture four spectral images using an image splitter with 542 nm, 558 nm, 586 nm and 605 nm band-pass filters. Before the spectral images can be used to analyze the optical density ratios they have to go through a preprocessing stage where they are registered together in software. The registration process is different from previous registration algorithms because the images are at very low resolution and therefore we propose a more data oriented method. To validate the registration we propose a method which uses the distance between edges in images. This method could be valuable to quantify the registration accuracy in more general image registration problems.

Original languageEnglish
Pages (from-to)5-8
Number of pages4
JournalReport - Helsinki University of Technology, Signal Processing Laboratory
Publication statusPublished - 2004
EventProceedings of the 6th Nordic Signal Processing Symposium, NORSIG 2004 - Espoo, Finland
Duration: 9 Jun 200411 Jun 2004


Dive into the research topics of 'Automatic registration of spectrophotometric retinal images'. Together they form a unique fingerprint.

Cite this