Recognition of Tamil Handwritten Characters using Daubechies Wavelet Transforms and Feed-Forward Backpropagation Network

MJose, Tiji; Wahi, Amitabh

doi:10.5120/10656-5422

Cited by 9 publications

(4 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For experimental purposes, the authors have made use of limited samples collected from different native Tamil speakers. Jose et al in [39] have collected samples from 100 users, with each class having 100 samples. There is no mention of how many classes are considered for data collection and experimentation.…”

Section: Figure 5 Sample Images From the Hpl Databasementioning

confidence: 99%

uTHCD: A New Benchmarking for Tamil Handwritten OCR

Shaffi¹,

Hajamohideen²

2021

IEEE Access

View full text Add to dashboard Cite

The robustness of a typical Handwritten character recognition system relies on the availability of comprehensive supervised data samples. There has been considerable work reported in the literature about creating the database for several Indic scripts, but the Tamil script has only one standardized database up to date. This paper presents the work done to create an exhaustive and extensive unconstrained Tamil Handwritten Character Database (uTHCD). The samples were generated from around 850 native Tamil volunteers including school-going kids, homemakers, university students, and faculty. The database consists of about 91000 samples with nearly 600 samples in each of 156 classes. This isolated character database is made publicly available as raw images and Hierarchical Data File (HDF) compressed file. The paper also presents several possible use cases of the proposed uTHCD database using Convolutional Neural Networks (CNN) to classify handwritten Tamil characters. Several experiments demonstrate that training on the proposed database helps traditional and contemporary classifiers perform on par or better than the existing dataset when tested with unseen data. With this database, we expect to set a new benchmark in Tamil handwritten character recognition and serve as a launchpad for developing robust language technologies for the Tamil script.

show abstract

Section: Figure 5 Sample Images From the Hpl Databasementioning

confidence: 99%

uTHCD: A New Benchmarking for Tamil Handwritten OCR

Shaffi¹,

Hajamohideen²

2021

IEEE Access

View full text Add to dashboard Cite

show abstract

“…Features to be given as input are derived from a set of measurements used for pattern recognition. There prevails a difficulty in representing the pattern of classification on the basis of structural information in this statistical approach [2]- [8].…”

Section: A Statistical Approachmentioning

confidence: 99%

Text Analysis and Information Retrieval of Historical Tamil Ancient Documents Using Machine Translation in Image Zoning

Vellingiriraj¹,

Balamurugan²,

Balasubramanie³

2016

IJLLL

View full text Add to dashboard Cite

“…Thereafter, the National Mission for Manuscripts has engaged in active research to preserve medical documents and so far, successfully digitized about 10,000 manuscripts (Joshi, 2016). Handwritten character identification applications employed three essential methods for digitizing historical records: the statistical (Nagy, et al 1992; Jose and Wahi 2013; Vellingiriraj et al 2016), structural or syntactic (Chen and Lopresti 2012; Pornpanomchai and Daveloh 2007), and neural network‐based approaches (Kruatrachue, Pantrakarn, & Siriboon, et al 2007; Sanguansat, et al 2004; Budsayaplakorn, Asdornwised, & Jitanpunkul, 2003).…”

Section: Introductionmentioning

confidence: 99%

“…three essential methods for digitizing historical records: the statistical (Nagy, et al 1992;Jose and Wahi 2013;Vellingiriraj et al 2016), structural or syntactic (Chen and Lopresti 2012;Pornpanomchai and Daveloh 2007), and neural network-based approaches (Kruatrachue, Pantrakarn, & Siriboon, et al 2007;Sanguansat, et al 2004;Budsayaplakorn, Asdornwised, & Jitanpunkul, 2003). Kumar, Sreekumar and Athvankar (2009) emphasized that the palm leaf practice is one of the oldest methods of language and classical encryption in India, especially in Southern India.…”

mentioning

confidence: 99%

Creation of original Tamil character dataset through segregation of ancient palm leaf manuscripts in medicine

Subramani

Subramaniam²

2020

Expert Systems

View full text Add to dashboard Cite

Palm leaves were one of the essential and primary sources for writing before the advent of paper. Tamil is one of the oldest southern Indian languages and among the ten oldest languages of the world. Agathiyar, a renowned Siddhar of ancient India, considered as the Father of Siddha Medicine, wrote all his therapeutic procedures only on palm leaf manuscripts, in the Tamil language. For modern day, people who try to only write and read new aspects of Tamil, identifying the ancient characters of the language is difficult. To expand readability and secure the written medicinal practices and traditions, a better recognition system is needed that can transform the ancient text images to modern ones, interpreting the ancient Tamil characters from palm leaves and understanding their contexts a time-consuming and complicated process. Especially when it comes to medicine, the practitioners need to understand the contents of a manuscript, to apply them on a daily basis. Therefore, a recognition system is of much use to understand, interpret, and apply the techniques explained in the manuscript on a daily basis. This study is an attempt to create a considerable volume of Tamil character datasets through the segregation of ancient Tamil palm leaf manuscripts related to the field of medicine. In this study, the characters created are fed as inputs to expert systems for intelligent recognition of the context and content perceived to be present in the selected medical manuscripts. The characters have been identified in large numbers manually, and datasets are created using Gaussian distortion. K E Y W O R D S ancient Tamil characters, expert systems for character recognition, Gaussian distortion, palm leaf manuscripts in Tamil, Tamil and medicine, Tamil medical manuscripts 1 | INTRODUCTION Tamil is one of the oldest and renowned languages in the entire world with a strong literature domain. In the prehistoric period, the poets, particularly in Tamil Nadu, used palm leaves to conceal information. The ancient literature integrates several palm leaf works that include Sangam literature, masterpieces, Saiva, Vaishnava, medical works, cuisine, astrology, Vaastu, jewels, songs, dance and drama, and Siddha (Vellingiriraj and Balasubramanie, 2013; Vellingiriraj and Balasubramanie, 2014). In the last decade, the importance of the ancient medical manuscripts in Tamil and attention towards the need to restore them have attracted the interest of many researchers. The preserved ancient medical manuscripts in Tamil by saints such as Agathiyar, were created for the first time, through a digitization process in a seminar hosted in 2006 by Foundation for Revitalisation of Local Health Traditions. Thereafter, the National Mission for Manuscripts has engaged in active research to preserve medical documents and so far, successfully digitized about 10,000 manuscripts (Joshi, 2016). Handwritten character identification applications employed

show abstract

Recognition of Tamil Handwritten Characters using Daubechies Wavelet Transforms and Feed-Forward Backpropagation Network

Cited by 9 publications

References 9 publications

uTHCD: A New Benchmarking for Tamil Handwritten OCR

uTHCD: A New Benchmarking for Tamil Handwritten OCR

Text Analysis and Information Retrieval of Historical Tamil Ancient Documents Using Machine Translation in Image Zoning

Creation of original Tamil character dataset through segregation of ancient palm leaf manuscripts in medicine

Contact Info

Product

Resources

About