In this paper, we present a novel approach for efficient and
high-quality surface registration on large and complex 3D
shapes based on the combination of local and
global geometric spatial information. In this approach,
at first, instead of relying on one type of scanned data, we
propose to use two types of scanning data provided that
it can support both global and local shape information.
The scanned low-resolution global 3D shape data supplies
the global shape structural prior for registering the
high-resolution local 3D surface patches. Local surface
patches can thus be optimally registered requiring less
overlapping and thus reducing redundancy.
Furthermore, due to the restriction of hardware,
we cannot directly process and register hundreds of scanned local surfaces to a global low-resolution shape model at one time. We segment the low-resolution global 3D shape model into several meaningful parts using a newly proposed  variational 3D shape segmentation algorithm. The multiple local surface patches can be registered on these segmented
parts respectively and all the segmented parts can
be merged after registration. To verify the feasibility
of the proposed approach, this approach has been evaluated
for acquiring various real biological 3D models. Using
only geometric spatial information from local surface
patches and global shape model without using texture
information, the results show that the proposed
approach can achieve efficient and high-fidelity
surface registration on large and complex 3D shape models.