Automatic Pupil Detection in Ophthalmology Diagnose

Automatic pupil detection is often required as a necessary function in modern Ophthalmology diagnosis. After analyzing the major challenges existing in various diagnosis environments, this paper suggests a stepwise work–flow combining machine vision algorithms including spatial convolution, circular Hough transform and region growing, to provide robust and precision alignment between diagnostic system and patient’s pupil. Experiments on a large scale image data–set as well as on live devices show that critical requirements in clinic diagnosis have been satisfied.