Recognition of On-line Handwritten Arabic Digits Using Structural Features and Transition Network

In this paper, an efficient structural approach for recognizing on-line handwritten digits is proposed. After reading the digit from the user, the coordinates (x, y) of the pixels representing the drawn digit are used for calculating and normalizing slope values of these coordinates. Successive slope values are then used to record the change of direction which used to estimate the slope. Based on the changing of signs of the slope values, the primitives are identified and extracted. These primitives represent a specific string which is a production of a certain grammar. Each digit can be described by a specific string. In order to identify the digit we have to determine to which grammar the string belongs. A Finite Transition Network which contains the grammars of the digits is used to match the primitives’ string with the corresponding digit to identify the digit. Finally, if there is any ambiguity, it will be resolved. The proposed method is tested on a sample of 3000 digits written by 100 different persons; each person wrote the 10 digits three times each. The method achieved accuracy of about 95% on the sample test. Experiments showed that this technique is flexible and can achieve high recognition accuracy for the shapes of the digits represented in this work.