Multiscale Approach of Retinal Blood Vessels Segmentation Based on Vessels Segmentation with Different Scales

摘要

In this work, we developed retinal blood vessels segmentation approach using contrast limited adaptive histogram equalization, morphological filtering, k-means clustering, matched filtering for thin and thick vessels selection. We also applied matched filtering for thin vessels selection using the kernels which were built in order to determine the existence of line segments with different length and orientation. Our approach has shown promising results.