Unicode Mizo Character Recognition System using Multilayer Neural Network Model
J. Hussain1, Lalthlamuana2
1Jamal Hussain, Department of Mathematics & Computer Science, Mizoram University, Aizawl, Mizoram, India.
2Mr. Lalthlamuana, Department of Mathematics & Computer Science, Mizoram University, Aizawl, Mizoram, India
Manuscript received on May 04, 2014. | Revised Manuscript received on May 04, 2014. | Manuscript published on May 05, 2014. | PP: 85-89 | Volume-4 Issue-2, May 2014. | Retrieval Number: B2208054214/2014©BEIESP
Open Access | Ethics and Policies | Cite
©The Authors. Published By: Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/
Abstract: The current investigation presents an algorithm and software to detect and recognize pre-printed mizo character symbol images. Four types of mizo fonts were under investigation namely – Arial, Tohoma, Cambria, and New Times Roman. The approach involves scanning the document, preprocessing, segmentation, feature extraction, classification & recognition and post processing. The multilayer perceptron neural network is used for classification and recognition algorithm which is simple and easy to implement for better results. In this work, Unicode encoding technique is applied for recognition of mizo characters as the ASCII code cannot represent all the mizo characters especially the characters with circumflex and dot at the bottom. The experimental results are quite satisfactory for implementation of mizo character recognition system.
Keywords: Character Recognition, Neural Network, Multi-Layer Perceptron, and Unicode.