Vector field distance transform: An implicit alternative representation for mesh fusionand filtering in the surface reconstruction process of scanned objects

for mesh processing. Raw data produced by 3D scanners need to be processed to
reconstruct the surface of scanned objects. Mesh filtering to reduce acquisition noise
introduced by the scanner and mesh fusion to integrate multiple scans of an object into a
complete description of the object are two important steps in the reconstruction process
of a scanned object. The scalar field distance transform (SDT) of a 3D mesh is a
volumetric representation of the mesh which implicitly describes the surface of a scanned
object. This alternative representation was previously used to perform mesh operations
such as mesh fusion. This presentation introduces a new implicit representation based on
the SDT to improve the accuracy of this alternative representation and to enable mesh
filtering using such representation. A new vector field distance transform (VDT) is
proposed and triangulation algorithms used to reconstruct the SDT resulting mesh are
adapted to the new VDT. The mesh fusion method applied to the SDT is also adapted to
the VDT to improve the results quality of this mesh operation using the distance
transform alternative representations. Two adaptive filtering methods applied on the VDT
are designed to reduce acquisition noise while preserving the meshes geometric features.
Mesh filtering and mesh fusion results using these new designs are shown and compared
to other methods to highlight the advantages of the proposed algorithms.