Automatic Polling Seeded Region Growing (APSRG) for Segmentation of Blood Vessels in Fundus

Seeded Region Growing (SRG) method can be applied to similar or extremely small image regions for avoiding over-segmentation with complex backgrounds. The original SRG is just using the Fuzzy method and combined with Region Growing to have the edge with a tiny region. In this research, SRG can be applied to the Blood vessel in Fundus. A blood vessel in Fundus is placed in the retina. This tiny blood vessel in Fundus can detect using SRG. However, this method has a problem to detect SRG from the background. The problem from SRG is still detecting the unwanted edge from the blood vessel in Fundus. This problem can be solved with Harris Corner edge detection method to avoid it. Based on that problem, this research proposed a novel algorithm Automatic Polling Seed Region Growing (APSRG) to automatically poll seed from SRG for image segmentation vessels in the fundus dataset. The Automatically poll seed was applied using Harris Corner edge detection. Harris corner edge detection is attached between the Fuzzy method and Region Growing from the SRG method. From that experiments, it was observed that from the averages Sensitivity has a value of 0.659, specificity has a value of 0.808 and the Accuracy has a value of 0.9565.