Up

Feature vector or time-series – comparison of gestures representations in automatic gesture recognition systems

Feature vector or time-series – comparison of gestures representations in automatic gesture recognition systems
File Size:
1.09 MB
Author:
Katarzyna Barczewska, Wioletta Wójtowicz, Tomasz Moszkowski
Email:
kbarczew[at]agh[dot]edu[dot]pl, tmoszkow[at]agh[dot]edu[dot]pl, wioletta[dot]wojtowicz[at]mech[dot]pk[dot]edu[dot]pl
Date:
04 May 2015
Downloads:
36 x

Abstract: In this paper, we performed recognition of isolated sign language gestures - obtained from Australian Sign Language Database (AUSLAN) – using statistics to reduce dimensionality and neural networks to recognize patterns. We designated a set of 70 signal features to represent each gesture as a feature vector instead of a time series, used principal component analysis (PCA) and independent component analysis (ICA) to reduce dimensionality and indicate the features most relevant for gesture detection. To classify the vectors a feedforward neural network was used. The resulting accuracy of detection ranged between 61 to 87%.

Keywords: principal component analysis (PCA), independent component analysis (ICA), Neural Networks, sign language, automatic recognition

Area: Biomedical Engineering

 
 

License Agreement

I agree to the terms listed above
 
 
Powered by Phoca Download