Image-to-image retrieval reflecting human sensibility and association

Image-to-Image retrieval (I2I) accompanied by data compression are presented. Given a query image, the presented retrieval system computes PCA and/or ICA bases by extracting source information. On the data compression in the sense of rate-distortion, this method outperforms JPEG which uses DCT. Besides, PCA and ICA bases reflect each image’s edge and texture information. Therefore, these bases can be used to define similarities combined with color information. Such similarity measures are tested to be compatible with human sensibility and association, or human “kansei”, via a set of opinion tests. Designed systems are proven to be effective and productive to such similar-image retrieval. In addition to static database, images collected through the Internet and WWW which are based on Text-to-Image retrieval (T2I) are tested Free annotations attached to such images are often irresponsible and erroneous. Therefore, filtering on so collected images is necessary. This paper’s 121 system can filter out such wrong images without sorting by human tasks. All source codes and object codes are open and can be downloaded form the authors’ web site.