A Novel Approach for Blood Vessel Edge Detection in Retinal Images

Blood vessel appearance is an important indicator for many diagnoses, including diabetes, hypertension, and arteriosclerosis. Blood Vessel edge detection in retinal images is very important in medical image processing. A lot of algorithms have been suggested for extracting medical image edges, and however, obtaining continuous edges with less over-detection points is difficult for edge extraction. In this paper, we propose a novel approach for blood vessel edge detection by integrating wavelet with fractal. First, accurate edges are extracted by employing wavelet technique. Then, continuous edges are obtained based on fractal technique. Finally, the experiments aptly show that our algorithm can obtain better edge continuity comparing with canny operator Chang (9) applied contextual-based Hopfield neural network for finding the edges of CT and MRI images. Gudmundsson et al. (10) developed an algorithm that detected well-localized, unfragmented, thin edges in medical images based on optimization of edge configurations using a genetic algorithm. Many applications in medical image processing need to extract the edge characteristic of image, hence, obtaining continuous edges is very important in medical image edge detection. In this paper, we propose a new approach to obtain continuous blood vessel edges by integrating wavelet with fractal. The main steps of our method are described as follows. First, the initial medical image edges are obtained using discrete wavelet transform. Then the continuous medical image edges are obtained based on fractal methods. II. INITIAL EDGE EXTRACTION Multiscale edges can be detected and characterized from the local maxima of a wavelet transform. Mallat et al. (11) obtained an excellent image edges based on a quadratic spline wavelet with compact support. In this paper, we use a spline wavelet to obtain the initial medical image edges, the details of algorithm are described as follows. First, we construct the spline wavelet and wavelet filter coefficients, then, obtain medical image edges based on discrete wavelet transform. Based on the method of (12), we obtain a discrete spline