Online Farsi Character Recognition Using Structural Features

Abstract

In this paper, grouping and recognition of online Farsi discrete characters are presented according to their structural features. The letters are divided into 9 groups based on the form and structure of their main bodies. After feature extraction, grouping is performed using a decision tree. Final recognition of letters is carried out in each group by delayed strokes. The proposed method is a rapid method in character recognition because time-consuming methods have not been used. Our proposed method was tested on TMU-OFS dataset, and a recognition rate of 94% and 92% was achieved for character grouping and recognition, respectively. The mean processing time for recognizing a letter was 3ms.

Keywords

• Recognition
• Grouping
• Online handwriting
• Decision tree
• Farsi
• Discrete letters