Iterative Vessel Segmentation with Stopping Criterion for Fundus Imagery

Vessel segmentation in fundus images plays vital role in diagnosing and treating patients in Ophthalmology. This proposed vessel segmentation algorithm consists of three stages to improve the lower contrast fundus images includes enhancement followed by thresholding and segmentation. Adaptive histogram equalization method is used to enhance the input image. From the enhanced image the major vessel are extracted by thresholding using gray thresh method. The new vessel pixels are identified iteratively using region growing method in which a new stopping criterion is introduced to improve the accuracy. The proposed method outperforms than the existing method of iterative vessel segmentation which achieves 3% greater in accuracy.