Analyses of Graphical Features of Textiles by Human Impressions

A picture database is being created by collecting textile pictures in order to aid designers from a graphical aspect. Classification of the pictures adopts eight scales ranging each from -3 to +3 based on eight pairs of adjectives in antonyms, such as warm/cool. Our research about the classification was carried out with the data of questionnaires by some thirty inspectors for about fifty sample pictures. An aim of our research is to classify more thousands of textile pictures by a computer eye, automatically. We propose graphical features which can represent human first impressions against given pictures. The graphical features are given by each value along the scales corresponding to the pairs of adjectives, such as warm/cool. We found the warm/cool scale could be automatically calculated from the image processing. The scale can become a powerful parameter to classify most textile pictures.