Correlation Analysis between Wood Eigen Textures and Perceptual Qualities

摘要

In this paper, we analyzed the relationship between woodrnEigen textures and the human perception. First, we developed arndatabase of one hundred wood textures. Then, through ourrnexperiment that was designed based on the Fleming’s subjectivernexperiment (Journal of Vision, 2013), nine types of perceptualrnquality scores were assigned to each texture. The perceptualrnqualities are glossiness, transparency, colorfulness, roughness,rnhardness, coldness, fragility, naturalness, and prettiness. Second,rnwe statically analyzed the one hundred wood textures using thernprincipal component analysis (PCA). The PCA bases of the woodrntexture database are called “wood Eigen textures.” Finally, wernobtained correlation coefficients between weights of wood Eigenrntextures and perceptual quality scores. Our results showed the firstrnwood Eigen texture (the first principal component) was correlatedrnwith the perceptual qualities of “glossiness” and “coldness.” Thernsecond one was correlated with the perceptual qualities of “coldness”rnand “colorfulness.” Based on the analysis, we demonstrate arntexture generation technique for arbitrary perceptual qualities.