A higher-order PDE-based image registration approach

✍ Scribed by Volker Grimm; Stefan Henn; Kristian Witsch

• Publisher: John Wiley and Sons
• Year: 2006
• Tongue: English
• Weight: 323 KB
• Volume: 13
• Category: Article
• ISSN: 1070-5325
• DOI: 10.1002/nla.467

✦ Synopsis

This paper addresses the problem of image registration with higher-order partial di erential equation (PDE) methods. From the study of existing a ne-linear and non-linear methods, a new framework is proposed that uniÿes common image registration methods within a generic formulation. Currently image registration strategies are classiÿed into either a ne-linear or non-linear methods subject to the underlying transformations. The new approach combines both strategies to obtain proper approximations which are invariant under global geometrical distortion (shearing), anisotropic resolution (scale changes), as well as rotation and translation. To achieve this favourable property, a modiÿed gradient ow approach is proposed which uses an operator with a kernel consisting of a ne-linear transformations. An approximation with ÿnite di erences leads to a large singular linear system. The pseudo-inverse solution of this system can be computed e ciently by augmenting the singular system to a regular system. Numerical experiments show the improvements compared to unmodiÿed gradient ow approaches.