Retinal Image Registration Using Intensity Values and Bifurcation Structure

Retinal image registration is important for applications such as diagnosis and treatment in ophthalmology system. The process by which obtaining correspondence between digital images of the same retina but in different viewpoints. Methods for the registrations are difficult because the variation in structure and intensities. Retina is the only part in human body where the blood vessels can be directly visualized. This method is combining the feature based method and intensity of the particular points in blood vessel structure. Considering the bifurcation structure formed by the blood vessels consisting of one main points and three neighbor points connected to it. Twelve angle values and three length components are used to find the characteristic vector of a single structure. Matching of these characteristic vectors of two images is more accurate and efficient than conventional point matching methods. Combining the intensity values along with the structure makes it more accurate hybrid method of retinal image registration.