Volume Image Registration
Arthur Ardeshir Goshtasby · 2017
This chapter first discusses methods for detecting feature points in volumetric images. Then, it details methods for determining homologous points in images are detailed. To account for local geometric differences between the given images, a coarse-to-fine search is taken. The coarse-to-fine search strategy not only speeds up the correspondence process but also reduces the number of outliers. Next, the chapter describes various transformation models for registration of volumetric images and given examples of volume image registration. Volume spline and weighted rigid transformations are used to register volumetric images. A volume spline transformation uses some global interpolation functions, while a weighted rigid transformation uses some locally sensitive approximation functions. The key performance measures in registration software are accuracy, reliability, and speed. The chapter further describes methods to determine these performance measures in volumetric image registration. Finally, the chapter reviews literature relating to volume image registration.