Gestures are spatiotemporal signals that contain valuable information. Humans can understand gestures with ease, but for computers or robots it is a challenging task involving thousands of computations per video frame. Current state of the art gesture recognition systems treat gestures as Markov Chains. Then the task of gesture recognition is to match the incoming video sequence to these Markov Chains. Each Markov State is modeled with spatial features such as hand location and temporal features like the motion vectors. The main problem with this approach is the high order of computational complexity. In this paper we propose a novel gesture recognition technique based on projecting the temporal axis information onto the spatial plane. Then this spatial intensity image is fed to a machine learning classifier (SVM in our case) for recognition. We show that the proposed algorithm achieves an accuracy that is comparable to the current state of the art approaches, but with a (much) reduced computational burden.

Using time proportionate intensity images with non-linear classifiers for hand gesture recognition / Ahmad, Omar; Khosa, Ikramullah; Bona, Basilio; Anjum, MUHAMMAD LATIF. - STAMPA. - (2013), pp. 1-12. (Intervento presentato al convegno The 8-th international conference on robotic, vision, signal processing & power applications tenutosi a Penang (Malaysia) nel 11th and 12th November 2013).

Using time proportionate intensity images with non-linear classifiers for hand gesture recognition

AHMAD, OMAR;KHOSA, IKRAMULLAH;BONA, Basilio;ANJUM, MUHAMMAD LATIF
2013

Abstract

Gestures are spatiotemporal signals that contain valuable information. Humans can understand gestures with ease, but for computers or robots it is a challenging task involving thousands of computations per video frame. Current state of the art gesture recognition systems treat gestures as Markov Chains. Then the task of gesture recognition is to match the incoming video sequence to these Markov Chains. Each Markov State is modeled with spatial features such as hand location and temporal features like the motion vectors. The main problem with this approach is the high order of computational complexity. In this paper we propose a novel gesture recognition technique based on projecting the temporal axis information onto the spatial plane. Then this spatial intensity image is fed to a machine learning classifier (SVM in our case) for recognition. We show that the proposed algorithm achieves an accuracy that is comparable to the current state of the art approaches, but with a (much) reduced computational burden.
File in questo prodotto:
Non ci sono file associati a questo prodotto.
Pubblicazioni consigliate

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11583/2520896
 Attenzione

Attenzione! I dati visualizzati non sono stati sottoposti a validazione da parte dell'ateneo