LAUSR

In this paper we propose the application of a novel associative classifier, the Heaviside's Classifier, for the early detection of Age-Related Macular Degeneration un retinal fundus images. Retinal fundus images are, first, processed by a simple method based on the Homomorphic filtering and some basic mathematical morphology operations; in the second phase we extract relevant features of the images using the Zernike moments, we also apply a feature selection method to select the best features from the original features set. The dataset created from the images with the best features are used to train and test a new classification model whose learning and classification phases are based on the Heaviside's Function. Experimental results show that our method is capable to achieve an accuracy value about the 94.12% with a dataset created from images belonging to famous image repositories.