Development of a low cost android application to detect human eyes, iris and pupil from video or image

Abstract

Smartphone is a skyrocketing product which gains popularity among all age groups, especially the younger generation. This may lead to serious and much earlier eye diseases such as macular degeneration or retinopathy. Pupil size can reveal the fatigue of a person. Research shows average pupil diameter changes according to level of sleepiness. Base on this, tiredness alert system may develop to prevent overuse of smartphone. Therefore, an accurate and low cost pupil size detection system in smartphone can be essential. Besides, since Android shares over three quarters of the smartphone market, development of Android application could be widely used. Thus, this project aims to develop a low-cost eye detection system in android application. The system can trace human eyes, irises and pupils through video stream or static images efficiently. The system first detects the face and the eyes area. Next, iris and sclera locations and radius are determined using fast iterations and image processing. Finally, the pupil center and radius are estimated using simple gradient-based methods. Evaluations are done using public datasets of faces and irises. Results show that the iris detection is precise and improved. Meanwhile, improvement of pupil size detection is still in progress.