The Key Technology Of Fundus Retinal Blood Vessels Segmentation Research

With the development of digital image processing technology, medical image processing and analysis is becoming the focus on hot issues in the field of image processing. Modern medicine and digital image processing technology increasingly close ties, greatly improving the accuracy of diagnosis by the doctor on the clinical cases. The severity of retinal microvascular changes closely related to cardiovascular and cerebrovascular diseases, diabetes, high blood pressure and eye diseases. So research effective segmentation algorithm of the retinal vascular network, and assist doctors to early diagnosis and treatment of these patients has a very important significance. Because the retinal vascular network in fundus depths is the only non-invasive direct observation of capillaries in the human body, use it as identification also has a high level of security. Based on the former researches on the retinal blood vessel segmentation algorithm, a combination algorithm of maximum entropy threshold and adaptive threshold is proposed. The algorithm consists of three steps. In the first step, the retinal image is enhanced by using an improved matched filter. Then the thin vessels are obtained by applying local entropy thresholding and the thick vessels are obtained by adaptive thresholding. Finally, the blood vessels are extracted by applying the logical OR operation on the detected thin and thick vessels. In addition, we also studied the fuzzy clustering method. Using an improved Fuzzy C-means clustering algorithm on the retinal image processing. Calculating the two-dimensional histogram of the enhanced image, and then take the diagonal elements of it, which are smaller influence by the noise. By using these elements to update the cluster centers, then achieve iteration. Finally, the blood vessels are extracted by the cluster matrix from the last iteration.