I. Introduction
Advances in information technology made it easier to conduct research on photo-visions. By using this technology, the visual cognition of the photographed object can be analyzed. The bottleneck in the research using photos is the quantification or tagging of photos. We quantified the photos by measuring their visual similarities as shown in Fig. 1. The inter-photo visual similarity (IVS) was measured by the percentage of people who felt that the two photos were “similar”. However, practically this method could not measure the IVS for a large number of photos because of huge cost in time and effort.