Chinese Coin Recognition Based on Unwrapped Image and Rotation Invariant Template Matching

Abstract

This paper proposes a Chinese coin recognition method based on unwrapped image and rotation invariant template matching. At first unwrap the reference coin and the specimen coin, which makes a rotation of the coins convert to a translation of the profile. Then lengthened the unwrapped specimen coin image and the unwrapped reference coin image always lies in the unwrapped specimen if they are the same class. Template matching is carried out for every allowed position of the template in the specimen image. Residual coins, which can not be differentiated by template matching directly, are refined by local template matching, YCbCr and inclination detection of two close line segments. The method was implemented to identify the Chinese coins and results were encouraging.