Hindi Handwritten Character Recognition Using Histogram of Oriented Gradients Feature

Handwritten Character Recognition is among one of the illustrious fields of research nowadays. The field of Pattern Recognition has a huge demand in handwritten character recognition of old historical documents available in different languages. Since in each language the characters have different shapes with different styles of writing which varies from user to user such as size, tilt, rotation, etc. Therefore, feature extraction is a key step which affects the accuracy or recognition of the characters. The feature selection depends upon the types of characters to be recognized. However, a lot of research has been carried out on recognition of different languages such as English, Chinese, Arabic, etc. But handwritten character recognition of some Devnagari languages such as Hindi and Sanskrit are still in their infant stages. Therefore, this area still demands a lot of research for their identification with high accuracy. In this paper, an approach for recognition of handwritten Hindi Characters is proposed using a global handcrafted feature descriptor named HOG. The first five characters of Hindi consonants are recognized using SVM classifier with an accuracy of 97.20%.