A Supramodal Vibrissa Tactile and Auditory Model for Texture Recognition

Audition and touch endow spectral processing abilities allowing texture recognition and discrimination. Rat whiskers sensory system exhibits, as the cochlea, resonance property decomposing the signal over frequencies. Moreover, there exists strong psychophysical and biological interactions between auditory and somatosensory corteces concerning texture analysis. Inspired by these similarities, this paper introduce a "supramodal" model allowing both vibrissa tactile and auditory texture recognition. Two gammatone based resonant filterbanks are used for cochlea and whiskers array modeling. Each filterbank is then linked to a feature extraction algorithm, inspired by data recorded in the rats barrel cortex, and finally to a multilayer perceptron. Results clearly show the ability of the model for texture recognition in both auditory and tactile tuning. Moreover, recent studies suggest that this resonance property plays a role in texture discrimination. Experiments presented here provide elements in the direction of this resonance hypothesis.