I've implemented ICP (Iterated Closest Point) to register the 3D data from different angles in order to fuse them into a entity.

Another application is: a 3D facial recognition algorithm was proposed, using ICP method.

A 1-1 test ( i.e. given two faces, it tells if they are the same person or not ) was carried out. The algorithm use 3D geometry, basically carrying out a ICP and based on the error to determine if it is a match.