Toward human motion search using fingerprinting

This paper investigates a 'fingerprinting' technique for describing human motion sequences. This work shows that human motion fingerprints can facilitate the search of human motion within large databases, similar to the fingerprinting approach used for the search of audio and image databases. This paper investigates the extraction of a reliable set of features from human motion capture data sequences that can be combined to generate a unique fingerprint. Results show that the fingerprints could be used to reliably differentiate between unique motions. © 2012 IEEE