Performance Analysis of Coin Matching Using Local Texture Features

At present coin recognition and classification tasks are performed manually by the numismatists. This process requires lot of time and proper preservation of hard copies of source documents such as printed catalog. Hence automation in this area is very much helpful and supportive. In this paper machine vision based automatic coin recognition using local texture features such as SIFT (Scale Invariant Feature Transform) and SURF (Speed Up Robust Features) algorithms are proposed to complement the conventionalmanual classification and recognition process. This work is carried out on ancient Roman coins and the performance is compared.