DFT based feature extraction technique for recognition of online handwritten Gurmukhi strokes
Keerti Aggarwal, Rajendra Kumar Sharma · 2016
This paper implements a feature extraction technique for recognizing online handwritten Gurmukhi characters. For attaining high recognition accuracy in such a system, computation of suitable features is an important task. DFT (Discrete Fourier Transform) based feature extraction technique is employed in this work. In this paper, we have considered 86 stroke classes of Gurmukhi script. We have taken 75-100 variations per class in the data set. To calculate the recognition accuracy, a data set of 8408 stroke samples has been considered. A recognition accuracy of 91.7% has been achieved when 11-fold cross-validation approach in LibSVM with RBF kernel is used.