PASCAL - Pattern Analysis, Statistical Modelling and Computational Learning

A Voting Margin Approach for the Detection of Retinal Microaneurysms
Ilkka Autio, Juan Carlos Borrás García, Ilkka Immonen, Petri Jalli and Esko Ukkonen
In: IASTED's 5th Conference on Visualization, Imaging and Image Processing (VIIP-2005), September 7-9, 2005, Benidorm, Spain.


We address the problem of detecting microaneurysms in retinal fundus images. Retinal microaneurysms are the earliest known indicator of diabetic retinopathy, an affliction which may result in blindness. We derive a maximum margin classifier capable of utilizing a collection of strong base classifiers, each of which may impose a different similarity induced kernel in the input space. Our experiments demonstrate that the resulting classifier is accurate and at least on par with the previous approaches to the problem.

Postscript - Requires a viewer, such as GhostView
EPrint Type:Conference or Workshop Item (Paper)
Additional Information:ML-based retinal microaneurysms detection whose performance is similar to the methods described by Walther et al. at ISMDA2002.
Project Keyword:Project Keyword UNSPECIFIED
Subjects:Machine Vision
Learning/Statistics & Optimisation
ID Code:1817
Deposited By:Juan Carlos Borrás García
Deposited On:29 November 2005