A comparison between the performances of several distances for isolated handwritten Arabic numerals recognition

In this paper we have achieved a comparison between the performances of different types of distances used in the k nearest neighbors classifier for recognition of isolated handwritten Arabic numerals extracted from Mnist database, more precisely these distances are those of Euclid, of Manhathan, of Minkowski, and of Tchebychev, to do this, we have used for pre-processing the image numerals the median filter, the thresholding, the centering and the skeletonization techniques and in order to extract the features from numerals we have employed the zoning method. The simulation results that we obtained demonstrates that all these distances are generally almost equivalents. © 2015 SERSC.