Research On Automatic Classification System Of Retinal Image For Cataract Detection Based On Expert Knowledge

Cataract is one of the most common diseases that might cause blindness. Previous research shows that cataract occupies almost50%in severe visually impairments. Considering the fact that retinal image is one of the most important medical references that help to diagnose the cataract. The classifier building procedure includes three parts: pre-processing, feature extraction, and classifier. In the pre-processing part, an improved top-bottom hat transformation is proposed to enhance the contrast between the foreground and the object, and a trilateral filter is used to decrease the noise in the image. The method of threshold and the algorithm of Otsu are used to segment the image respectively. There are two kinds of method to complete the work. One is mainly based on the texture message of the image whose classifier is the neural network classifier; the other is based on the PCA plus LDA whose classifier is the minimum distance classifier. Based on the clearness degree of the retinal image, the patients’cataracts are classified into normal, mild, medium or severe ones. The initial evaluation results illustrate the effectiveness of our proposed approach; their correct rates can be up to80%.