Eye is one of the vital sensory organs in the human physiological system. It supports visual recognition by converting light-signal into neural signals with the help of a thin tissue layer; known as retina (Almazroa et al., 2015; Adalarasan and Malathi, 2018). Normal functioning of the eye is always necessary to sense the external data available in the form of the light/picture. The regular performance of the eye is forever preferred to supply the light based sensory signal to the brain. Vision can be disturbed due to various reasons, such as eye injury, inflammation, diabetes, aging and other physical illness (Burlina et al., 2017; Choi et al., 2017; Bokhari et al., 2018).
1.1 Problem Statement
The vision interruption due to aging is a severe difficulty. Early recognition and treatment can help to maintain sharp eye vision. The Age-related Macular Degeneration (AMD) is a universal retinal disease among the aged humans (>50 years) (Kanagasingam et al., 2014; Fauw et al., 2017). The earlier works in the literature confirms that, AMD is the major cause of vision loss; which initially affects the macula in eye and slowly decreases the vision. Based on its severity, this AMD is categorized as (i) early, (ii) intermediate, and (iii) late phase. If the detection and treatment of the AMD is initiated in its first (early) phase, then it can be treated easily with various procedures, such as drug injection, laser treatment and surgery (Mookiah et al., 2013; 2014).
Due to its importance, a number of research works are proposed by the researchers to detect the AMD using benchmark and clinical grade Digital-Fundus-Retinal-Images (DFRI) (Ramachandra et al., 2015; Qureshi et al., 2016). Recent works confirms the availability of various machine-learning (ML) and Deep-Learning (DL) procedures (Koh et al., 2017; 2018; 2019; Ting et al., 2019) to detect AMD with greater accuracy. Most of the methods proposed discuss the relation of the retinal Optic-Disc (OD) and Retinal-Cup (RC) ratio (Yang et al., 2018), RD extraction and evaluation (Tan et al., 2017), and image feature extraction using the gray/RGB scale DFRI and its classification (Tan et al., 2016; Meier, M.H. et al., 2016). Each of the work has its own merits and demerits.