Loading...
Automatic optic disc detection from low contrast retinal images
Author(s)
Wisaeng, Kittipol
Hiransakolwong, Nualsawat
Pothiruk, Ekkarat
Date Issued
October 16, 2012
Type
Article
Abstract
On the research work leading to automatic analysis from low contrast retinal images the knowledge of optic disc (OD) detection is very essential, and a novel method to automatically locate OD is proposed in this paper. The method starts by following key preprocessing step, i.e., color normalization, contrast enhancement and removal noise. The OD detection algorithm is based on combination of mathematical morphology and Otsu's algorithms are presented. The proposed method was evaluated using the publicly available of the STARE project's dataset, containing 81 retinal images of both normal and diseased retinal. The OD was detected correctly in 91.35%. In addition, the OD was detected correctly in 97.61% using the local dataset.
Citation
Applied Mathematical Sciences, 6(101-104), 5127-5136, 2012