Retinal vessel segmentation is an essential phase of diabetes detection. In detecting retinal vessels, image processing and machine learning play a vital role. Machine learning algorithms enhance the segmentation ratio of retinal images. This paper proposed machine learning-based retinal vessel segmentation methods. The proposed methods use wavelet transform methods to extract texture features and support vector machine applied for retinal vessel detection. The proposed algorithm improves the segmentation ratio of the retinal vessel. The lower content of the retinal image increases the range of the feature sample and maximizes the cover of the support vector. The proposed algorithm is simulated in MATLAB tools—the DRIVE dataset used to test the proposed algorithm.
Key words: Segmentation, Retinal Image, SWT, SVM, CNN.
|
