Categorization of Persian Detached Handwritten Letters Using Intelligent Combinations of Classifiers

Detecting optical characters is considered as the main responsibility to convert printed documents and manuscripts to digital format. In this article, detecting Persian handwritten letters using the combination of classifiers and features are assessed employing geometric and statistical sections' features. In order to detect each letter, it is divided into two parts; the major and the minor parts. Then, some features are presented for them. Preprocess algorithm prepares the possibility to unify dimension features for multiple words and delivers them to classifier for detection. The hierarchy classification can be obtained by separating the letters. In the following, the optimal answer will be reached by using GA method of different SVM, ML and KNN classifications. Extraction algorithm of required features is proved by using the evaluation of the basis of PCA. Empirical results represent classification of 94.3 and 92 accuracy in simple and multiple parts in 20 times repetition, respectively.