In this study, a method based on level sets has been proposed for improving performance in segmenting of blood vessels in the retina images. Because the accurate segmentation performance of blood vessels poses a significant role in the diagnosis and treatment of eye diseases. First, in the proposed method, noise is removed from the input image and the structure of the blurred image is sharpened due to noise elimination. Then the image intensity is homogenized and the structure information of the image is obtained. In the last step, blood vessels are segmented by an efficient level set method. In terms of quantity and quality, the efficient segmentation of blood vessels with the proposed method has also been proven by experimental results.