Boosting Gabor Feature Extraction for Gesture Recognition using Tri-axis Accelerometer

In this paper, a new feature extraction method for gesture recognition based on single tri-axis accelerometer has been proposed. We present a novel human computer interaction for cell phone through recognizing seventeen complex gestures. First of all, we propose 1D Gabor coefficients of acceleration signals as features for gesture recognition. Although Gabor wavelets can extract discriminative information from acceleration signals effectively, dimensionality of the feature space is very high. To cope with this problem, we adopt boosting algorithm to select and compress the Gabor feature. The Classifier we used is Multi-class Support Vector Machine. Experimental results demonstrate that the proposed method which not only can solve the problem but also can improve recognition accuracy. The encouraging results indicate that gesture recognition based on single tri-axis accelerometer can provide a novel human computer interaction.