Skip to Main Content
The authors propose to extract local texture features for image-based coin recognition in this study. A set of Gabor wavelets and local binary pattern (LBP) operator are employed to represent texture information. Concentric ring structure is used to divide the coin image into a number of small sections. Statistics of Gabor coefficients or LBP values within each section is then concatenated into a feature vector to represent the image. A circular shift operator is proposed to make Gabor features robust against rotation variance. Matching between two coin images is done via distance measurement. The nearest-neighbour classifier is used to classify a given test coin. The public MUSCLE database consisting of over 10 000 images is used to test our algorithms; results show that significant improvements over edge distance-based methods have been achieved. The authors have also analysed the performance of the system on recognising unregistered coins and the analysis suggests further improvement could be achieved if physical properties like diameter and thickness are included for feature representation.