A method of style discrimination of oil painting based on 3D range data

We conducted an experiment about "Kansei", which is sensibility of each person. Our goal was to extract an artist's style in the oil painting. A style is determined by color skill and other key elements. Our methods paid particular attention to the characteristics of touch, one of the skills. Texture of the oil painting is very different from watercolor painting because paints harden, and it becomes a ruggedness that can easily be seen by the human eye. We supposed that a touch could be determined with in the local mean curvature and the Gaussian curvature extracted from the range image of the oil painting. In this paper we proposed a style discrimination method using these curvatures, and reported on the experiment result.