Extracting Appearance Information inside the Pupil for Cataract Screening

In this paper, we propose a method for cataract screening. Referring to the fact of increasing number of cataract especially for the people who lived in the developing countries, this kind of discussion will be important. The method used three kinds of features; specular reflection appearance, texture uniformity and average intensity inside the pupil. Actually, The core of our method is based on a specular reflection appearance inside the pupil. Referring to the reflection theorem, normal eye images will have two kinds of reflection inside the pupil while cataract eye images have one reflection and both of them are always in a line so we can easily distinguish between cataract and normal eye images. However, some times we face the problems of distinguishing specular reflection and whitish color caused by serious cataracts. It will cause a misclassification that a serious cataract are regarded as a normal. In this paper we propose an improved method for handling such kind of problems by extracting statistical texture consists from an uniformity and an average intensity inside the pupil. The result shows that percentage value of True Positive is 91.97% and percentage value of False Positive is 18.75%. This performance is good enough for cataract screening especially in the developing countries.