Recognition of Online Handwritten Bangla Characters Using Supervised and Unsupervised Learning Approaches

Recognition of Online Handwritten Bangla Characters Using Supervised and Unsupervised Learning Approaches

Prosenjit Mukherjee (Future Institute of Engineering and Management, India), Shibaprasad Sen (Future Institute of Engineering and Management, India), Kaushik Roy (West Bengal State University, India) and Ram Sarkar (Jadavpur University, India)
Copyright: © 2020 |Pages: 13
DOI: 10.4018/IJCVIP.2020070102
OnDemand PDF Download:
No Current Special Offers


This paper explores the domain of online handwritten Bangla character recognition by stroke-based approach. The component strokes of a character sample are recognized firstly and then characters are constructed from the recognized strokes. In the current experiment, strokes are recognized by both supervised and unsupervised approaches. To estimate the features, images of all the component strokes are superimposed. A mean structure has been generated from this superimposed image. Euclidian distances between pixel points of a stroke sample and mean stroke structure are considered as features. For unsupervised approach, K-means clustering algorithm has been used whereas six popular classifiers have been used for supervised approach. The proposed feature vector has been evaluated on 10,000-character database and achieved 90.69% and 97.22% stroke recognition accuracy in unsupervised (using K-means clustering) and supervised way (using MLP [multilayer perceptron] classifier). This paper also discusses about merit and demerits of unsupervised and supervised classification approaches.
Article Preview

1. Introduction

Handwriting recognition is a hot research topic due to its numerous applications since long. Few of such applications include reading postal addresses, bank check amounts, and retrieval of data from filled-in forms, signature verification and so on. Handwriting recognition can be done in one of the two approaches: online and offline. In offline recognition, the handwritten documents are scanned and then the scanned images are used fed to the recognition system. In contrary, online recognition system recognizes the writing when user writes; i.e. in real time.

An advantage of online data capturing device with writing facility is that it stores temporal or dynamic information of the writing. This information consists of the number of strokes, order of the strokes, direction of the writing for each stroke, and the speed of the writing within each stroke. This extra information makes online handwriting recognition a bit easier than offline handwriting recognition. It is to be noted that a stroke is the writing from pen down to pen-up.

Research works on online handwriting recognition is gaining more popularity due to enhanced usage of the devices like tab, smart phones etc. In these devices, people can provide information as freely as they are habituated to write with age-old pen and paper. Here, not only the chances of mistyping become less but also the information is supplied in lesser time in comparison to handling with a keyboard.

For both offline and online modes, handwriting recognition can be performed in both supervised and unsupervised approaches. In supervised approach, the extracted features from the character samples are assigned proper class labels to train the model by a classifier. An unknown sample is then predicted by using the trained model. In case of unsupervised technique, class labels are not used, rather different clustering algorithms are used to make the groups representing different pattern classes. An unknown character sample is then predicted by computing closest distance from the cluster centroids.

Efforts made by different researchers for the recognition (both supervised and unsupervised approaches) of handwritten characters in different scripts are discussed in this section. Authors in (Sajjad et al., 2013) have shown character recognition scheme on MINIST dataset by using K-means clustering technique under unsupervised learning approach. Authors in (Mathur & Saroliya, 2014; Gaur & Yadav, 2015) have experimented on Hindi characters using clustering approach. The experiments mentioned in (Mathur & Saroliya, 2014; Gaur & Yadav, 2015; Gaur & Yadav, 2015) are performed on offline databases. There are lots of works on online character recognition for different scripts in supervised learning way. Some good research contributions for online handwritten Devanagari and Gurumukhi scripts are described in (Connell et al., 2000; Joshi et al., 2005; Bawa & Rani, 2011; Santosh et al., 2012; Santosh et al., 2015; Tripathi et al., 2012; Kubatur et al., 2012; Lajish & Kopparapu, 2014). Few research investigations towards online handwritten Bangla character recognition are mentioned below.

Complete Article List

Search this Journal:
Volume 13: 1 Issue (2023): Forthcoming, Available for Pre-Order
Volume 12: 4 Issues (2022): 1 Released, 3 Forthcoming
Volume 11: 4 Issues (2021)
Volume 10: 4 Issues (2020)
Volume 9: 4 Issues (2019)
Volume 8: 4 Issues (2018)
Volume 7: 4 Issues (2017)
Volume 6: 2 Issues (2016)
Volume 5: 2 Issues (2015)
Volume 4: 2 Issues (2014)
Volume 3: 4 Issues (2013)
Volume 2: 4 Issues (2012)
Volume 1: 4 Issues (2011)
View Complete Journal Contents Listing