Post-processing of 3D scanned data is still the bottleneck for a wider diffusion of this technology. In this pages we describe our second generation tools for processing 3D scanned data. In particular, our tools support: range maps alignment, range maps merge (or fusion), mesh simplification and color attribute management. The software suite has been implemented by scratch and encompasses both up-to-date solutions and some original methods (merging, simplification, color management and, in part, alignment).
The architecture of the software suite is described and an evaluation of its use in the framework of a complex acquisition in the Cultural Heritage domain (3D scanning of a bronze statue) is reported.

Scanning an object or an architectural complex entails executing a set of rather complex tasks, which are generally called 3D scanning pipeline:

• acquisition planning; decide the set of range maps (how many, view specification of each of them) to be taken to obtain a complete sampling of the object's surface;

• scanning the artefact from different viewpoints, producing a set of range maps;

• range maps alignment; by definition, range map geometry is relative to the current sensor location and has to be transformed into a a common coordinate space where all the range maps lies well aligned; after alignment, the sections of the range maps which correspond to the same surface zone will be geometrically overlapping;

• range maps merge (or fusion); build a single, non redundant mesh out of the many, partially overlapping range maps;

• mesh editing; improve (if possible) the quality of the reconstructed mesh;

• mesh simplification; reduce the huge complexity of the model obtained, producing different level Of Details (LOD) representations;

• and finally, map to the surface mesh the surface attribute data (e.g. color or BRDF sampling).

We have designed a suite of software tools that manages all of the above phases (excluding the acquisition planning phase, onto which we are now working):

• MeshAlign v.2: the module allows the registration of multiple range maps; it adopts a classical approach based on a pairwise local and global alignment, implemented with a number of innovations to reduce the user contribution, to improve efficiency and easy of use, and finally to support the management of a large number of range maps (we processed up to six hundreds range maps).

• MeshMerge: the module allows the reconstruction of a single 3D mesh out of a set of registered range maps. Two different approaches were implemented: a classical volumetric reconstruction approaches based on distance field and a new approach characterized by a lower space complexity, higher efficiency and improved accuracy with respect to the previous, but more sensible to the alignment residual error.

• MeshEdit: the module allows to perform simple editing actions on the mesh(e.g. to fill small holes, to remove non-manifold components of dangling edges/faces, to apply smoothing filters, etc).

• MeshSimplify: the module supports the simplification of the [huge] meshes produced by 3D scanning devices, by removing mesh vertices in a controlled manner. The simplification follows the edge collapse approach and has been implemented in an out-of-core fashion to allow the management of meshes that could be larger than the core memory of the computer used.

• Weaver: the module supports the management of a set of images (either produced by the scanner, or taken with a digital camera) and builds up a texture map wrapped around the 3D model.

Our 3D scanning tools have been used in many different complex acquisition projects. Among them, we present here some results obtained while scanning the Minerva of Arezzo, a bronze statue (155 cm.) of the Museo Archeologico (Florence, Italy) now under restoration.
Four different models of the Minerva have been scanned in 2000-2002. The following Table presents data on: scanning system used, software used, processing time and output data complexity. The Minerva experience is a very good example to assess the evolution of our post-processing tools. The first acquisition was performed on Oct. 2000 with the first generation of our tools. The following two acquisitions were done using progressively improved versions, while the fourth scan was performed using an alpha version of the second generation tools

M. Callieri, P. Cignoni, F. Ganovelli, C. Montani, P. Pingi, R. Scopigno
VCLab's Tools for 3D range data processing
4th International Symposium on Virtual Reality, Archaeology and Intelligent Cultural Heritage (VAST2003) and First EUROGRAPHICS Workshop on Graphics and Cultural Heritage, Brighton (UK), 5-7 November 2003

M. Callieri, P. Cignoni, and R. Scopigno
Reconstructing Textured Meshes from Multiple Range RGB Maps
7th Int.l Fall Workshop on Vision, Modeling, and Visualization 2002, Erlangen (D), Nov. 20 - 22, 2002

P.Cignoni, C.Rocchini, C.Montani, R.Scopigno
External Memory Management and Simplification of Huge Meshes
IEEE Trans. on Visualization and Comp. Graphics (to appear)

P. Cignoni, C. Montani, and R. Scopigno
A Comparison of Mesh Simplification Algorithms.
Computers & Graphics, Pergamon Press, Vol. 22(1), 1998, pp. 37-54.

C. Rocchini, P. Cignoni, F. Ganovelli, C. Montani, P. Pingi and R. Scopigno
Marching Intersections: an Efficient Resampling  Algorithm for Surface Management
International Conference on Shape Modeling and Applications (SMI 2001)
Genova, Italy, 7-11 May 2001.

C.Rocchini, P.Cignoni, C.Montani, P.Pingi, R.Scopigno
A Low Cost Scanner Based on Structured Light
Computer Graphics Forum  (Eurographics 2001 Conf. Proc.), vol. 20 (3), 2001, 299-308. 2001

C.Rocchini,P.Cignoni,C.Montani,P.Pingi,R.Scopigno
A Suite of Tools for the Management of 3D Scanned Data
3D Digital Imaging and Modeling Applications of: Heritage, Industry, Medicine & Land,
Workshop Proc., Padua (I), 3-4 April, 2001