Registration of retinal images using blood vessel features

The registration of retinal images has significant importance in the operations of additional diagnosis and to determine the ailment progression of various retinal diseases. This paper deals with a feature based technique for retinal image registration. The purpose of the proposed method is to assist ophthalmologist to remark variations that happen in the blood vessel structure because of different pathological abnormalities. The retinal image comprises of various aspects such as optic disc(OD), blood vessels(BV), macula, fovea etc. In our approach blood vessels are taken as main feature as blood vessels can be considered as an important benchmark for disorders viz. diabetes, age related macular edema and retinopathy of prematurity (ROP). The segmentation of blood vessels is done by using morphological operation. For the registration section, the OD is segmented from which the OD center is calculated. The center of the OD is considered as control point and it is used to correct the orientation difference between the two images to be registered. By minimizing the rotational and translational differences between them, the registration is achieved.