Tactile Rendering of Fabric Textures Based on Texture Recognition

Abstract

A tactile rendering system for fabric textures is presented to improve the tactile sensations in humancomputer interactions in this paper. A tactile platform based on electrovibration is constructed to produce the virtual tactile sensation. The voltage amplitude perception threshold and differential threshold of human beings on the system are measured by a series of experiments. On this basis, the relationship between the stimulus signal and tactile perception is established using psychophysical experiments. A texture library of fabrics is then created by using a surface morphology analyzer. The amplitude of the corresponding voltage for rendering textures can be calculated according to the established relationship between the stimulus signal and tactile perception. As a new fabric texture image is input to the system, convolutional neural network (CNN) is used to match the texture with the texture library and then the corresponding stimulus signal can be generated. The above system is finally evaluated and numerically quantified through experiments.