Rahele Allahverdi - Qazvin Islamic Azad University, Qazvin, Iran
Azam Bastanfard - Karaj Islamic Azad University, Karaj, Iran

In a situation in which a pattern recognition system intends to be run in a real world environment, access to an adequate database is highly demanded in view of the fact that this database has the capability of improving overall system during three major phases namely training, validating, and testing. One of the novel and challenging trends in the field of statistical pattern recognition is ancient coins classification which suffers from accessing standard image databases. To the best of our knowledge, related studies which are conducted up to date do not have published their database. Our focus in this paper is not only on promoting Sassanid coins image database with accessible and standard data, but also on presenting the unique application of this database in testing algorithm which are dealt with imbalanced data. This database, which we refer to as SaCo DB, contains 2076 images from both sides of 1038 coins. These coins belong to Sassanid era. Essential demographic data is also provided in SaCo with the aim of facilitating the process of images.

image database, ancient coins, Sasanian coins, middle Persian, imbalanced database