Automatic Cataract Detection in Fundus Retinal Images using Singular Value Decomposition

Abstract

Cataract is a considerable age-related eye disease. Many people across the world are losing their vision because of cataract. The only possible treatment is painful surgery. So, non-invasive automatic computer-aided techniques are developed to detect and diagnose the cataract at earlier stages. In this paper, a new automatic cataract detection (ACD) method is proposed to detect the cataract from digital fundus images. The detection of cataract is a two class problem that can classify the input fundus images into normal and cataract. The proposed ACD method uses first twenty singular values from singular value decomposition as features. Then these features are applied to the support vector machine for classification. The percentage of testing accuracy (detection accuracy) obtained for cataract detection is 97.78. From the obtained results, it is clear that, the proposed method outperforms with the existing methods in evaluating the cataract.