Diabetic Retinopathy Classification Using HARALICK Features
Reshma Sultana1, K.S.Rajasekhar2

1Reshma Sultana, Department of ECE, ANU college of Engineering and Technology, “Centre of Excellence in VLSI Design & Antennas”, Acharaya Nagarjuna University, Guntur (Andhra Pradesh), India.
2K.S Rajasekhar, Department of ECE, ANU college of Engineering and Technology, Acharaya Nagarjuna University, Guntur, (Andhra Pradesh), India.

Manuscript received on 23 March 2019 | Revised Manuscript received on 30 March 2019 | Manuscript published on 30 March 2019 | PP: 1323-1327 | Volume-7 Issue-6, March 2019 | Retrieval Number: F2902037619/19©BEIESP
Open Access | Ethics and Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)

Abstract: Diabetes is one of the most rapidly growing health threat around the world. Diabetic Retinopathy is abnormalities due to diabetes that affects eyes which leads to blindness over a period of time if not detected and cured in early stages. So detection and classification of Diabetic Retinopathy in early stages is important. In this paper, work is done on two database’s one is DIARETDB0 database and other one is HRF (High-Resolution Fundus) database. The DIARETDB0 database consists of total 130 color fundus images, among which 20 are normal fundus images and 110 fundus images are having signs of Diabetic Retinopathy. The HRF (High-Resolution Fundus) database consists of total 45 color fundus images, among which 15 are normal fundus images, 15 fundus images are having signs of Diabetic Retinopathy and 15 fundus images are of glaucomatous patients. The texture features are extracted using Haralick Feature extraction Process. The Haralick Features are nothing but combination of Haar-DWT (Discrete Wavelet Transform) features and GLCM (Gray-level co-occurrence matrix) features that is original image is decomposed using Haar-DWT then sub-band images are produced. These sub-band images are used to extract features by GLCM. The proposed algorithm for feature extraction gives promising results. The Haralick Features are used to classify the normal and diabetic retinopathic images using a classifier. The performance of classification is calculated using the term Accuracy.
Keywords: Diabetic Retinopathy, Fundus images, Haralick Features, Classification, Accuracy.

Scope of the Article: Classification