This invention is directed to a highly accurate and efficient method and algorithm, namely the Dual-Bootstrap Iterative Closest Point algorithm, for performing image registration generally and retinal image registration in particular. Retinal image registration is challenging. The images are projections of a curved surface taken from a wide range of viewpoints using an un-calibrated camera. The non-vascular surface of the retina is homogeneous in healthy retinas, and exhibits a variety of pathologies in unhealthy retinas. Unfortunately, for the purposes of registration, these pathologies can appear and disappear over time, making them poor choices for conventional longitudinal registration. Only the vasculature covers the entire retina and is relatively stable over time. Recognizing these limitations, this invention is directed to a novel retinal registration approach using localized vascular structures that include both the vessels themselves and their branching cross-over points. This software overcomes technical hurdles associated with traditional registration methods by using mathematical algorithms to initialize registration, achieve minimization by avoiding local minima caused by misalignments between vessels, and robust enough to account for missing structures.