In this paper we describe the pattern recognition methods used for the identification of coins in the new coin recognition and sorting system Dagobert developed at the ARC Seibersdorf research centre. The purpose of Dagobert is to sort high volumes of coins. In particular the coin detection task, the features used to pre-select meaningful potential master coins for the verification process and the verification/rejection method itself will be detailed. Some recognition results are given at the end of the paper, indicating that the approach fulfils the given task effectively.