F.1 – A Simple Algorithm for Automatic Alignment of Ocular Fundus Images

Publisher Summary This chapter describes an automatic alignment algorithm for registration of ocular fundus images. In order to enhance vessel structures, a spatially oriented bank of filters designed to match the properties of the objects of interest is used. To evaluate interframe misalignment a fast cross-correlation algorithm is adopted. The performances of the method have been estimated by simulating misalignments between image pairs. When there are more than two images to be registered, the accuracy of the registration method can be quantified using a cross-validation approach. As concerne image sequence integration, the adopted non-linear filtering allows a reasonable trade-off between computational complexity and detection sensitivity/specificity. The alignment algorithm is fully automatic. Despite its simplicity, the method is accurate enough for the image sequences dealt with, and it is faster in calculations as compared to other sequential methods. It can be useful in clinical applications such as the study of retinal blood flow and the analysis of capillary-network morphology.