Isotropic Undecimated Wavelet Transform Fuzzy Algorithm for Retinal Blood Vessel Segmentation

The segmentation of retinal blood vessels is of significance in retinal image analysis but has some technical difficulties since its easy-to-get noise effect during imaging. In this study, we proposed an isotropic undecimated wavelet transform (IUWT) fuzzy algorithm for retinal blood vessel segmentation. After image preprocessing, we utilized IUWT to denoise the retinal image in its frequency field, and then we performed a robust fuzzy clustering algorithm on texture features based on local gray value entropy to segment retinal blood vessels automatically. Post-processing morphological treatments were performed to refine the segmentation results. Two public datasets were used to test our segmentation method. The sensitivity, specificity and area under receiver operating characteristic curve was 0.8205, 0.9018 and 0.9375 respectively for our proposed method, with time cost about 13.4 s in a personal computer, which is comparable with other algorithms. Therefore, we believed that our proposed algorithm is an available method that could be applied in segmentation for retinal image analysis.