Proceedings of 2018 ASEE Northeast Section Conference

A Multi-Anatomical Retinal Structure Segmentation System for Automatic Eye Screening Using Morphological Adaptive Fuzzy Thresholding
jasem almotiri

Intensity inhomogeneity is one of the major challenges for image segmentation in general and for retinal image segmentation in particular. Besides, using one single algorithm for identifying and extracting multi-anatomical structures is considered the most challengeable duty in retinal image segmentation due to the dramatic difference between different retinal anatomical and non-anatomical structures. In this poster, A novel, robust, and stand-alone algorithm based on hybrid combination between adaptive fuzzy thresholding and mathematical morphology used to extract multi-anatomical objects with broad features and characteristics in retinal image is proposed. Retinal Vessels, Optic Disk and Exudate lesions are extracted without the need for texture analysis or synthesis. The algorithm was validated using three benchmark databases. In addition, it performs well at extracting vessels and optic disc from pathological retinal images.

Last modified: 2018-04-09
Full Text: DOCX

<< Back to Proceedings