Inviscid Model for Nonrigid Image Registration

(Co-author: Zhao Yi)

A mathematical model is proposed for nonrigid image registration. The solution of a set of nonlinear partial differential equations gives the spatial transformation for registering two images. This model can accommodate both small and large deformations, with sharper edges and clear texture achieved at less computational cost. The performance of this model has been tested on a variety of images including 2D and 3D, mono-modal and multi-modal, synthetic and clinical data.

Registering a square to a circle

Sq2Cir.avi

2D segmented brain

template image target image

our model
fluid model
deformed template difference image before registration edge map

3D segmented coronal

A multiresolution technique is used for the 3D image registration.

slice 36
slice 39
slice 54
slice 60
template coarse deformed template fine deformed template target