Handwritten Kannada Numerals Recognition by Using Zone Features and CNN Classifier

Handwritten Kannada Numerals Recognition by Using Zone Features and CNN Classifier

Vishweshwarayya C. Hallur (Bharathiar University, Coimbatore, India), Rajendra S. Hegadi (Indian Institute of Information Technology, Dharawad, India) and Ravindra S. Hegadi (Solapur University, Solapur, India)
Copyright: © 2019 |Pages: 17
DOI: 10.4018/IJTHI.2019100106

Abstract

The proposed system presents a pre-processing, segmentation, features extraction approach and Deep Convolutional Neural Network (DCNN) classifier for recognition of handwritten Kannada numerals. Pre-processing have different steps like median filter, gray scale to binary, normalization, thinning, skew correction and slant removal. Segmentation process contains different methods like vertical projection profile for word and novel character segmentation. Collections of best discriminable features are very important part in achieving high rate of identification in automatic numeral detection systems. Kannada is the major south Indian character verbal by about 50 million people. This article presents a well-organized and novel technique for recognition of handwritten Kannada numerals using zone and distance matrix. An appropriate feature extractor and a superior classifier play most important task in achieving high detection rate for a recognition scheme. This article determines a variety of feature extraction approaches and classification techniques which are designed to recognize handwritten numerals of Kannada script. The DCNN classifier approach is used to classify the testing samples of each Kannada handwritten numerals. The experimental result gives the acceptable performance rate.
Article Preview
Top

2. Literature Survey

Kamble et al. (2015) proposed the Rectangle Histogram Oriented Gradient representation as the source for extraction of character. These algorithms need a small number of easy arithmetic operations per image pixel that formulates them appropriate for real-time applications. The dataset includes of 8000 models each one of 40 essential handwritten Marathi characters. Expected results by using classification methods are Support Vector Machines (SVM) and feed-forward Artificial Neural Network (FFANN) are presented.

Keysers and Daniel (2017), proposed a Google’s online handwritten identification scheme that presently supports 22 characters and 97 languages. The scheme centre of attention is on fast, high-accuracy text access for mobile, touch-enabled devices. They make use of an arrangement of state-of-the-art apparatus and merge them with novel additions in a flexible frame. This construction allows them to simply transmit developments involving languages and characters. They present a general summary of the scheme architecture and the new apparatus, like unified time- and position-based input interpretation, trainable segmentation, minimum-error rate training for feature extraction, and a cascade of pruning strategies, by achieving the experimental results for different setups.

Chaudhuri and Chandranath (2017) proposed the recognition and processing of struck-out texts in unrestricted offline handwritten file images. A feature extraction method is-based on two-classes, first one is SVM classifier utilized to detect moderate-sized struck-out parameters. The skeleton of the text element is measured as a graph and the strike-out stroke is recognized by using a constrained shortest path algorithm. They selected only 500 pages of documents and achieved an overall F-Measure of 91.56% (91.06%) in English script for struck out text detection.

Hegadi (2012) discussed an approach for recognition of handwritten as well as printed Kannada numerals. The planned framework involves the three main segments, such as pre-processing, feature extraction and classification. The numeral images are preprocessed and noise removal technique is applied, in the classification stage, a multilayer feed forward neural network classified ten handwritten numerals into distinct classes. The recognition rate of proposed method is impressive.

Complete Article List

Search this Journal:
Reset
Open Access Articles: Forthcoming
Volume 16: 4 Issues (2020): 2 Released, 2 Forthcoming
Volume 15: 4 Issues (2019)
Volume 14: 4 Issues (2018)
Volume 13: 4 Issues (2017)
Volume 12: 4 Issues (2016)
Volume 11: 4 Issues (2015)
Volume 10: 4 Issues (2014)
Volume 9: 4 Issues (2013)
Volume 8: 4 Issues (2012)
Volume 7: 4 Issues (2011)
Volume 6: 4 Issues (2010)
Volume 5: 4 Issues (2009)
Volume 4: 4 Issues (2008)
Volume 3: 4 Issues (2007)
Volume 2: 4 Issues (2006)
Volume 1: 4 Issues (2005)
View Complete Journal Contents Listing