Isolated printed Arabic character recognition using KNN and random forest tree classifiers

Abstract

Classification step is one of the most important tasks in any recognition system. This step depends greatly on the quality and efficiency of the extracted features, which in turn determines the efficient and appropriate classifier for each system. This study is an investigation of using both K- Nearest Neighbor (KNN) and Random Forest Tree (RFT) classifiers with previously tested statistical features. These features are independent of the fonts and size of the characters. First, a binarization procedure has been performed on the input characters images, and then the main features have been extracted. The features used in this paper are statistical features calculated on the shapes of characters. A comparison between KNN and RFT classifiers has been evaluated. RFT found to be better than KNN by more than 11 % recognition rate. The effect of different parameters of these classifiers has also been tested, as well as the effect of noisy characters.