Locally rigid globally non-rigid surface registration

We present a novel non-rigid surface registration method that achieves high accuracy and matches characteristic fea-tures without manual intervention. The key insight is to con-sider the entire shape as a collection of local structures that individually undergo rigid transformations to collectively deform the global structure. We realize this locally rigid but globally non-rigid surface registration with a newly de-rived dual-grid Free-form Deformation (FFD) framework. We first represent the source and target shapes with their signed distance fields (SDF). We then superimpose a sam-pling grid onto a conventional FFD grid that is dual to the control points. Each control point is then iteratively trans-lated by a rigid transformation that minimizes the difference between two SDFs within the corresponding sampling re-gion. The translated control points then interpolate the em-bedding space within the FFD grid and determine the over-all deformation. The experimental results clearly demon-strate that our method is capable of overcoming the diffi-culty of preserving and matching local features. 1