Optic disc segmentation based on Otsu's thresholding and level set

Retinal image inspection is crucial to identify and supervise a class of retinal diseases. Image processing procedures are widely considered to extract the infected region of retinal image in order to have a clear idea about the disease. In this paper, heuristic algorithm assisted multi-level thresholding and level set approaches are considered to extract the optic disc from the retinal image dataset. The proposed approach has three processing sections, such as (i) Heuristic algorithm guided multi-level thresholding, (ii) K-means clustering based separation and (iii) Level set based segmentation. The proposed approach is implemented on the RIM-ONE dataset and validated using the image similarity index values, such as Jaccard, Dice, False Positive (FP) rate and False Negative (FN) rate. The investigational outcome illustrate that, the proposed approach offers better values of image similarity index. Hence, the proposed technique is very significant for the segmentation and analysis of optic disc images.