A Highly Accurate and Robust Retinal Vessel Segmentation Algorithm

Retinal vessel segmentation is beneficial for eye surgery and detection of diabetic retinopathy. Mathematical morphology, or top-hat reconstruction can keep desired vessel structures. Gaussian mixture model is used here to build classification model and generate binary images of retinal vessels. After experimental testing, this work achieves the best vessel tracking ability and robustness performance