EEL 6825 Project Description: Coin Recognition

    For our project, we propose to design an image processing system to recognize the tail side of standard US coins using Mathematica and possibly C/C++. In other words we hope to be able to differentiate between the bald eagle on the quarter, the torch of liberty on the dime, Thomas Jefferson's house on the nickel, and the Lincoln Memorial on the penny. Images of the coins will be obtained by digitally photographing various coins on a solid colored background, such as a green piece of construction paper. The desire is to make the orientation of the image on the coin and camera distance from the coin irrelevant, assuming satisfactory detail can still be extracted. In order to make coin distance from the camera irrelevant, all coins will be normalized to the same size and only the images imprinted on the coins will be used for classification. Various forms of feature extraction are being considered including the use of edge detection, Fourier transforms, wavelets, and passing filters over the image for texture extraction. After feature extraction has been performed, LBG Vector Quantization will be used to locate coins in a feature vector space. When a coin image is processed and located in feature space, it will be matched up to a Vector Quantization region. Using Bayes Classifier the coin will then be classified as a quarter, dime, nickel or penny.

    Depending on time restraints, we also hope to implement more complexity and robustness to the coin recognition system. Additional work may include any of the following:

-Discrimination between coins and "not coins"
-Recognition of the head side of coins
-Recognition of multiple coins in one picture
-Recognition of overlapping coins

Works Consulted:

Project Home