This work introduces novel methods in the field of deformable image registration, both classical and learning-based. A central theme of the work is leveraging Reproducing Kernel Hilbert Space theory as a powerful regularization framework for dense displacement fields that are estimated via image registration.