Helder J Araujo, Rufino Silva, Jose F. Abreu, and Jose G. Cunha-Vaz
In this paper we described work currently under progress that aims at the 3D reconstruction of human retinas. Two modalities of images are employed: anglographic and laser-scan ophthalmoscopic. The specific nature of the images raises a number of difficulties because of the almost non-existent sharp boundaries. We show that some of these difficulties can be overcome by using algorithms that do not make explicit use of edges or other geometrical features.