Abstract— The purpose of this paper is to extract features from retina digital images based on a further analysis of high frequency components (HH) obtained with the discrete wavelet transform (DWT). In particular, the DWT is applied to the retina photograph to obtain its high-high (HH) image subband. Then, a further decomposition by DWT is applied to the HH image subband of the previous step to obtain HH*. Finally, statistical features are computed from HH*. The support vector machines (SVM) are employed to classify normal versus abnormal images using leave-one-out cross-validation method (LOOM). The simulation results show strong evidence of the effectiveness of features extracted from HH* than features extracted from HH. Thus, they are in accordance with our previous work where our approach was applied to mammograms. In summary, our methodology based on a further analysis of high frequency images using DWT helps extracting suitable features for automatic classification of normal and abnormal retina digital images.
Index Terms— retina digital image, discrete wavelet transform, high frequency subband, features extraction, support vector machines, classification
Cite: Salim Lahmiri, "Features Extraction from High Frequency Domain for Retina Digital Images Classification," Journal of Advances in Information Technology, Vol. 4, No. 4, pp. 194-198, November, 2013.doi:10.4304/jait.4.4.194-198
Copyright © 2013-2020. JAIT. All Rights Reserved
This work is licensed under the Creative Commons Attribution License (CC BY-NC-ND 4.0)