Segmentation and machine learning techniques applied to automatic detection of diseases from eye-fundus retina images

Abstract

Here are presented some methods of retina image processing for retina image segmentation of blood vessels, using the ARIA and B-COSFIRE filters methods, with noise reduction. Eye-fundus images are used for experiments for noise removal with KSVD and BM3D methods also we use an image contrast enhancement (CLAHE) method: Contrast limited adaptive histogram equalization. Is presented also an efficient method of image segmentation based on a nature inspired optimization technique namely Particle Swarm Optimization.. In the end machine learning techniques are applied to automatic detection of diseases from retina eye-fundus images with automatic classification.