Publication Date:
2015
abstract:
We present a very fast and simple-to-implement algorithm for the automatic registration of a large number of range maps. The proposed algorithm exploits a compact and GPU-friendly descriptor specifically designed for the alignment of this type of data. This pairwise registration algorithm, which also includes a simple mechanism to avoid to get false positives, is part of a system capable to align a sequence of up to hundreds of range maps in few minutes. In order to reduce the number of pairs to align in the case of unordered range maps we use a prioritization strategy based on the fast computation of the correlation between range maps through FFT. The proposed system does not need any user input and it was tested successfully on a large variety of datasets coming from real acquisition campaigns.
Iris type:
01.01 Articolo in rivista
Keywords:
Range Maps Registration; 3D Scanning; Automatic Registration
List of contributors:
Ganovelli, Fabio; Pingi, Paolo; Corsini, Massimiliano; Scopigno, Roberto
Published in: