Abstract— In this paper we describe a method for fast

and reliable coin classification using coin images taken in various

conditions by different people and a possible solution when

dealing with illumination inconsistencies. It relies on edge based

features and texture/colour features extracted automatically. For

edge detection we are using the Canny’s edge detector algorithm

with dynamic thresholds values obtained using the Otsu’s

threshold algorithm. At the end of this paper some experimental

results are given indicating that the approach fulfils the given

task effectively.