Persian handwritten digits recognition using machine learning algorithms

Optical character recognition (OCR) consists of three main stages: Preprocessing, feature extraction, and classification. The aim of preprocess section, is noise reduction, smoothing and normalizing the input data that can have vital role in distinguishing pattern space of feature. in the feature extraction section which considered as the most important section, every model is given a feature vector which is an introducer of that in the targeting feature space, sample and distinguishing from other samples further, feature extraction has significant effect on classification of sample class. The classification stage seeks to properly separate and classify feature vectors into different classes. One of the branches of pattern recognition is the recognition of Persian handwritten digit recognition. This paper introduces a method of Persian handwritten digit recognition, whose framework encompasses all three aforementioned stages: preprocessing, feature extraction, and classification.

کلیدواژه ها:

Persian handwritten digits recognition ، preprocessing ، feature extraction ، classification ، ensemble classifier