A Retinal Blood Vessel Segmentation Approach Based on Top-hat Transformation

The analysis of retinal blood vessel plays an important role in aided diagnosis, segmenting blood from retinal fundus images is the first operation. This paper proposes a novel approach for retinal blood vessel automatic segmentation. First, the top-hat transformation is used for blood vessel enhancement. Second, using double-threshold to segment the vessels from the enhanced fundus image, the high threshold is used for main vessel segmentation, and the low threshold is used for finer vessel segmentation. Third, fusing the finer blood vessels into main blood vessels. The experiment is tested in public fundus image database STARE. The result of the experiment shows that the method obtains good visual effect.