Research Articles
Recognition of handwritten characters using deep convolution neural network
Authors:
S. Arivazhagan,
Mepco Schlenk Engineering College, IN
About S.
Professor, Department of Electronics and Communication Engineering
M. Arun ,
Mepco Schlenk Engineering College, IN
About M.
Asst Professor, Department of Electronics and Communication Engineering
D. Rathina
Renganayagi Varatharaj College of Engineering, IN
About D.
Department of Electronics and Communication Engineering
Abstract
Character recognition is a very interesting technique in the field of pattern recognition. Specifically, handwritten character recognition is gaining the attention of researchers as it is necessary for historical documents, archives, or mass digitization of hand-filled forms. The correct classification of handwritten characters is really a challenging task due to its variability in the writing styles of an individual at different times and among different individuals such as size, shape, speed of writing and thickness of characters, etc. To solve this challenging task, the features extracted from the characters should be suitable for the variability of the characters. In this research work, Deep Convolution Neural Network (DCNN) has been used instead of hand-crafted features from the handwritten characters, to automatically learn the best features for this task. The proposed DCNN framework is trained and tested on the Chars74K handwritten dataset in all the aspects of handwritten numbers and handwritten English alphabets with various training and testing proportions and various subproblems. The recognition rate for the proposed DCNN provides better results when compared with the other schemes. The recognition rates for 62 classes in the Chars74K dataset with 50:50, 70:30 and 80:20 train test ratio is 88.05 %, 89.21 % and 90.32 %, respectively.
How to Cite:
Arivazhagan, S., Arun, M. and Rathina, D., 2021. Recognition of handwritten characters using deep convolution neural network. Journal of the National Science Foundation of Sri Lanka, 49(4), pp.503–511. DOI: http://doi.org/10.4038/jnsfsr.v49i4.9825
Published on
31 Dec 2021.
Peer Reviewed
Downloads
