抄録
In our previous work, to extract the KANSEI information received from a landscape image, we investigated the evaluation experiment by the Semantic Differential method. By analysing the obtained data using the factor analysis, we found out 4 factors associated with KANSEI information. In this paper, we consider the KANSEI modeling which is connecting with a landscape image and the KANSEI factors. By using MPEG-7 descriptors such as Scalable Color Descriptor and Edge Histgram Descriptor, the Saliency Map information of Visual Attention, and the KANSEI factors of monochromatic color, we can estimate the KANSEI information using the potential interpolation method. In addition, as an example of image database system applying this KANSEI modeling method, we consider the image retrieval system which used the KANSEI information as the reference key.
