DF-AOT: Dual Filter and Adaptive Otsu Thresholding for Retina Vessel Segmentation

Retinal vessel segmentation is crucial in analyzing retinal images for disease diagnosis. This paper proposes DF-AOT, an innovative unsupervised methodology for retinal vessel segmentation. DF-AOT aims to accurately segment blood vessels from retinal images, a crucial step in analyzing these images for disease diagnosis. The method leverages a unique combination of techniques to achieve this goal. The core concept revolves around a dual-filter strategy implemented throughout the processing pipeline. This strategy promises enhanced vessel segmentation compared to existing unsupervised and supervised methods. The manuscript will delve into the specifics of this dual-filter approach and its role in pre-processing, adaptive Otsu's thresholding, and post-processing stages. The effectiveness of DF-AOT will be evaluated on publicly available retinal image datasets, with comparisons drawn to previous methods to solidify its contribution to the field.