Abstract
In the design of product surface texture, it is important for a designer to grasp the effects that come from the difference between visual prediction and tactile experience. Such difference may affect the tactile evaluation, e.g. disappointment and exceeding expectation. In this paper, we propose a method for evaluating vision based prediction of tactile sensation. In the method, we ask participants to predict a tactile sensation of a texture surface by looking and judge a degree of similarity with another invisible texture surface by touching. With the judgments, we obtain subjective probability distributions of visually predicted touch on tactile dimensions. To demonstrate the effectiveness of the method, we applied the method to surface roughness of plastic crimp textures. Furthermore, we found that the entropy of brightness histogram as an image feature significantly correlates with visually predicted surface roughness of a plastic crimp texture.
