Abstract: Pterygium tends to affect those who are exposed to prolonged uv radiation in which Asian people are more susceptible to develop such condition. Thus, a computerized method that can automatically detect pterygium is greatly needed and is being proposed using image processing approach. The main focuses of this researchers is corneal segmentation of eye region in the Anterior Segment Photographed Images (ASPI). A heuristic approach based on Particle Swarm Optimization (PSO) thresholding is explored to segment the corneal to further improve the segmentation rate of non-ideal images that contain pterygium condition. To obtain the corneal after applying PSO, the three steps thresholding method is used based on frame differencing between multi color channels (HSV). Brazil Pterygium (BP) database is used to represent the pterygium condition while UBIRIS.V1, UBIRIS.V2 and Miles databases are used to represent the normal eye condition. Both pterygium and normal ASPIs have been tested rigorously to validate the robustness of the proposed algorithm. The results showed that BP have the highest accuracy (94%) while UBIRIS.V1 have the highest accuracy (93.2%) for the non pterygium database. In conclusion, it shows that the proposed method is mostly suitable for application in Asian countries in which there is increasing trend of pterygium cases.
Siti Raihanah Abdani, W. Mimi Diyana W. Zaki, Aini Hussain and Mohamad Hanif Md. Saad, 2016. Particle Swarm Optimization-Based Thresholding for Corneal Segmentation in Pterygium Detection. Asian Journal of Information Technology, 15: 1845-1850.