Authors : Mojtaba Mohammadpour, Abbas Mehdizadeh and Havva Alizadeh NoughabiAlizadeh Noughabi
Abstract: Handwritten digit recognition has a special importance and also different applications such as detecting handwritten addresses, cheques and documents. Persian handwritten digit classification is facing more difficulties due to different handwritten types and also inter-class similarities and intra-class differences. In this study, a method for Persian handwritten digits detection is proposing. In the proposed method, a mixture of Histogram of Oriented Gradients (HOG), 4-side profiles of the digit image and some horizontal and vertical samples of it is used. Then, the dimension of the feature vector is reduced by using Principle Component Analysis (PCA). In the classification step, Support Vector Machine (SVM) is used. The proposed method is applied on the HODA database. The results shows 99.25% in detection accuracy which is an adequate rate due to existing unacceptable samples in the database as well as achieving great improvement comparing to other existing methods.
Mojtaba Mohammadpour, Abbas Mehdizadeh and Havva Alizadeh NoughabiAlizadeh Noughabi, 2019. Persian Handwritten Digit Recognition using Support Vector Machine. International Journal of Signal System Control and Engineering Application, 12: 35-39.