Regularization Methods in Non-Rigid Registration :II. Isotropic Energies, Filters and Splines

The goal of this report is to propose some new regularization and filtering techniques specific to vector fields. Indeed, most of vectorial regularization energies used in the domain of image or feature matching are in fact scalar energies applied independently on each component of the transformation. The only common exception is the elastic energy, which enables cross-effects between components. In this report, we first propose a technique to find all isotropic differential quadratic forms (IDQF) of any order on vector fields, and give results in the case of order 1 and 2. The dense quadratic approximation induced by these energies give birth to a new class of vector filters, which are applied in practice in the Fourier domain. We also propose a family of isotropic and separable vector filters that generalize scalar Gaussian filtering to vectors and enables efficient isotropic smoothing without using Fourier transform, using recursive filtering in the real domain. We also study the splines induced by these energies in the context of spare point motion interpolation or approximation. These splines are generalization to vectors of the scalar Laplacian splines, such as the thin-plate spline. We finally propose to merge dense and sparse approximation problems, yielding solutions that mix convolution filter and splines. This enables to introduce naturally sparse geometric constraints into an intensity-based registration algorithm.

[1]  Hervé Delingette,et al.  Automatic Detection and Segmentation of Evolving Processes in 3D Medical Images: Application to Multiple Sclerosis , 1999, IPMI.

[2]  P. Kannappan,et al.  Rotation invariant separable functions are Gaussian , 1992 .

[3]  Daniel Rueckert,et al.  Nonrigid registration using free-form deformations: application to breast MR images , 1999, IEEE Transactions on Medical Imaging.

[4]  Colin Studholme,et al.  An overlap invariant entropy measure of 3D medical image alignment , 1999, Pattern Recognit..

[5]  Patrick Pérez,et al.  Medical Image Registration with Robust Multigrid Techniques , 1999, MICCAI.

[6]  A. Khotanzad,et al.  A physics-based coordinate transformation for 3-D image matching , 1997, IEEE Transactions on Medical Imaging.

[7]  H. Jeffreys On isotropic tensors , 1973 .

[8]  Karl Rohr,et al.  Approximating Thin-Plate Splines for Elastic Registration: Integration of Landmark Errors and Orientation Attributes , 1999, IPMI.

[9]  G. Marchal,et al.  Multi-modal volume registration by maximization of mutual information , 1997 .

[10]  Fabrice Heitz,et al.  Robust voxel similarity metrics for the registration of dissimilar single and multimodal images , 1999, Pattern Recognit..

[11]  Amir A. Amini,et al.  Comparison of Land-Mark-Based and Curve-Based Thin-Plate Warps for Analysis of Left-Ventricular Motion from Tagged MRI , 1999, MICCAI.

[12]  X. Pennec,et al.  3D non-rigid registration by gradient descent on a Gaussian-windowed similarity measure using convolutions , 2000, Proceedings IEEE Workshop on Mathematical Methods in Biomedical Image Analysis. MMBIA-2000 (Cat. No.PR00737).

[13]  Demetri Terzopoulos,et al.  Regularization of Inverse Visual Problems Involving Discontinuities , 1986, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[14]  R. Deriche Recursively Implementing the Gaussian and its Derivatives , 1993 .

[15]  M I Miller,et al.  Mathematical textbook of deformable neuroanatomies. , 1993, Proceedings of the National Academy of Sciences of the United States of America.

[16]  O. Faugeras,et al.  A variational approach to multi-modal image matching , 2001, Proceedings IEEE Workshop on Variational and Level Set Methods in Computer Vision.

[17]  Ron Kikinis,et al.  3D Image Matching Using a Finite Element Based Elastic Deformation Model , 1999, MICCAI.

[18]  Anthony J. Yezzi,et al.  A Curve Evolution Approach to Medical Image Magnification via the Mumford-Shah Functional , 2000, MICCAI.

[19]  R. Lakes Foam Structures with a Negative Poisson's Ratio , 1987, Science.

[20]  Nicholas Ayache,et al.  Regularization in Image Non-Rigid Registration: I. Trade-off between Smoothness and Intensity Similarity , 2001 .

[21]  Guy Marchal,et al.  Automated multi-modality image registration based on information theory , 1995 .

[22]  Jean-Francois Mangin,et al.  Multisubject Non-rigid Registration of Brain MRI Using Intensity and Geometric Features , 2001, MICCAI.

[23]  Jürgen Weese,et al.  Gray-Value Based Registration of CT and MR Images by Maximization of Local Correlation , 1999, MICCAI.

[24]  Christian Barillot,et al.  Bayesian approach to the brain image matching problem , 1995, Medical Imaging.

[25]  Michael Brady,et al.  Rotationally symmetric operators for surface interpolation , 1983, Comput. Vis. Graph. Image Process..

[26]  H. Weyl The Classical Groups , 1940 .

[27]  N. Ayache,et al.  Multimodal Image Registration by Maximization of the Correlation Ratio , 1998 .

[28]  David J. Hawkes,et al.  Voxel similarity measures for 3-D serial MR brain image registration , 1999, IEEE Transactions on Medical Imaging.

[29]  R. Deriche,et al.  Regularization and Scale Space , 1994 .

[30]  Jean Duchon,et al.  Interpolation des fonctions de deux variables suivant le principe de la flexion des plaques minces , 1976 .