The Importance of Magnitude Spectra in the Discrimination of Visual and Tactile Textures

Surface texture is one of the key parameters in the affective design of products. It can give tactile sensations which influence customers' appreciation and preference to the product. Among other tactile sensations, perceptual roughness is the most important. However, designing a surface texture which satisfies the aimed tactile-sensation is still challenging. To characterize roughness perception, roughness parameters have been used and they were found to not reflect the perceived roughness. In this research, instead of using roughness parameters, the surface contours were represented by using image textures. This method was motivated by two evidences; first, there is a strong indication that visual and tactile perception is related in some degree; second, the researches in image recognition are more mature and the knowledge in that field can be useful to model the tactile perception. There are at least four types of image features; statistical, structural, model-based methods, and spectral-based features. In this study, the FT spectra-based features were selected because they are more related to the frequency of textures. Furthermore, as there are two parts of FT spectra, this research was aimed to determine the importance of the magnitude and phase spectra of images in the discrimination of visual and tactile textures. In order to do so, some sets of tactile stimuli were designed by transforming pixels' value of image into the height of stimuli's asperities. The designs which are in 3D CAD files were then printed using a 3D printer. Four set of experiments were made; the first set was to investigate the similarity between visual and tactile perception in discriminating irregular textures; the second set was to investigate the influence of magnitude and phase spectra on the hybrid image appearance; the third set was to measure the magnitude dominance of images using power-spectra based features; the fourth set was to investigate the presence of natural textures which are magnitude dominant. The results show that the magnitudes are more important than the phases and the magnitude-based features can be used to measure the magnitude and phase dominance of textures. Beside that, the results also show that both visual and tactile perception have a similar pattern which indicates that it is possible to model the tactile perception using image textures.