An Autoencoding Method for Detecting Counterfeit Coins

We use coins in our daily life to pay for bus, metro tickets, vending machines, etc. However, the market antique and historical is another place, where quality of their genuinity play a significant role. Hence, researchers have considered different methods coin detection studies. In recent years 2-D 3-D image processing approaches been widely used image-based detection. this paper, we propose method detect counterfeit based on content. employed SIFT, SURF, MSER determine similarity degree datasets. Then, evaluate those descriptors by statistical analysis see which one most effective criterion According experiments, SIFT was selected as reliable algorithm Danish dataset. train an autoencoder find anomalies images. The trained receives input generates new image. output compared with basic using criterion. If between these two images meets threshold then genuine. Most require fake data training. This can be eliminated autoencoding-based anomaly method.