Gesture recognition using a NMF-based representation of motion-traces extracted from depth silhouettes

Aymeric Masurelle (1, 2) , Slim Essid (1, 2) , Gael Richard (1, 2)

1 LTCI - Laboratoire Traitement et Communication de l'Information
2 TSI - Département Traitement du Signal et des Images

We present a novel approach that classifies full-body human gestures using original spatio-temporal features obtained by applying non-negative matrix factorisation (NMF) to an extended depth silhouette representation. This extended representation, the motion-trace representation, incorporates temporal dimensions as it is built by superimposition of consecutive depth silhouettes. From this representation, a dictionary of local motion features is learned using NMF. Thus the projection of these local motion feature components on the incoming motion-traces results in a compact spatio-temporal feature representation. Those new features are then exploited using hidden Markov models for gesture recognition. Our experiments on a gesture dataset show that our approach is superior to more traditional methods that uses pose features and/or decomposition techniques such as principal component analysis, by taking advantages of an efficient NMF representation based on local spatio-temporal features.

Mots clés: Gesture recognition; Depth-silhouette; Motion-trace; Non-negative matrix factorisation; Hidden Markov models

Domaines: Traitement du signal et de l'image [eess.SP]; Traitement du signal et de l'image [eess.SP]

Aymeric Masurelle, Slim Essid, Gael Richard. Gesture recognition using a NMF-based representation of motion-traces extracted from depth silhouettes. 2014 IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP 14), May 2014, Florence, Italy. ⟨hal-00990252⟩