A Bilingual (Gurmukhi-Roman) Online Handwriting Identification and Recognition System
Gurpreet Singh1, Manoj Kumar Sachan2
1Gurpreet Singh, Department of Computer Science Engineering, CU, Gharuan
2Manoj Kumar, Sachan Department of Computer Science Engineering, SLIET Longowal
Manuscript received on 21 April 2019 | Revised Manuscript received on 26 May 2019 | Manuscript published on 30 May 2019 | PP: 2936-2952 | Volume-8 Issue-1, May 2019 | Retrieval Number: A1341058119/19©BEIESP
Open Access | Ethics and Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. 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: Advances in recent technologies and the power of Artificial Intelligence (AI) always motivate the researchers to work for Human-Computer Interaction (HCI). So that, HCI behaves similar to HHI (Human-Human Interaction). Various applications which come under HCI are automatic emotion detection from facial images and videos, sentiment analysis from textual data or abbreviation slangs, handwriting recognition etc. Automatic recognition of handwriting is one of the most complex task for computer systems. The factors like the existence of variety of languages with their unique scripts, difference among the writing styles of different writers or even same writer over time, delayed strokes, noise, requirement of huge amount of training etc. make this task most difficult. In this paper, the problem of Online Handwriting Recognition (OHR) is addressed for the text containing bilingual data. The input data is written by using English and Punjabi languages, so both Roman and Gurmukhi scripts are used for the purpose of writing. During feature extraction process, the stress is given to the local features by considering direction of strokes, position or order of strokes, slope and area covered by one complete stroke. Global features include the information about Headline stroke, straight lines and dots. Statistical features consider Zone identification and Direction code histogram. The final recognition is done by using Multi Layered-Perceptron (MLP) neural network as a classifier in the presence of two simultaneous recognition engines for Roman and Gurmukhi scripts respectively. Experimental observations show good results for the bilingual text.
Keywords: Bilingual Data; HCI (Human-Computer Interaction); HHI (Human-Human Interaction); Online Handwriting Recognition (OHR); Multi-Layered Perceptron (MLP) Neural Network
Scope of the Article: Pattern Recognition