Statictics of Gabor features for coin recognition

We present an image based approach for coin classification. Gabor wavelets are used to extract features for local texture representation. To achieve rotation-invariance, concentric ring structure is used to divide the coin image into a number of small sections. Statistics of Gabor coefficients within each section is then concatenated into a feature vector for whole image representation. Matching between two coin images are done via Euclidean distance measurement and the nearest neighbor classifier. The public MUSCLE database consisting of over 10,000 images is used to test our algorithm, results show that significant improvements over edge distance based methods have been achieved.