Coins Recognition Using Complex Moments and Feed Forward Neural Network

Abstract

In this research work, the set of complex moments is used as a feature vector to represent the contents of some coins cutouts. The extracted moment vector is fed to a trained feed forward neural network for recognition purpose. The feed back propagation algorithm was used to train the network. To improve the recognition performance some image processing stages (like, region of interest extraction, contrast enhancement) were applied before the determination of complex moments. The test results indicated that the use of complex moments leads to detection rate 71.9%, with 17.6% false alarms.K