Handwritten Kannada numerals recognition using deep learning convolution neural network (DCNN) classifier
- PDF / 2,703,982 Bytes
- 15 Pages / 595.276 x 790.866 pts Page_size
- 80 Downloads / 185 Views
ORIGINAL RESEARCH
Handwritten Kannada numerals recognition using deep learning convolution neural network (DCNN) classifier Vishweshwrayya C. Hallur1 • R. S. Hegadi2
Received: 11 July 2017 / Accepted: 2 April 2020 CSI Publications 2020
Abstract In pattern recognition, identifying Kannada handwritten numerals are a complex knot. This paper portrays an avenue that pikes us to attain a most potent Kannada Numerals recognition process. In this, handwritten Kannada characters are captivated in document fashion and are subjected to Pre-processing and attribute extraction processes. Pre-processing entails steps like noise removal, binarization, normalization, skew amendment, and thinning. Features are extricated by exploiting strategies like Drift Length Count, Direction related progression code, DWT and Curvelet Transfiguration Wrapping. For an impressive classification process deep convolution neural network classifier is preferred. Isolation accuracy of Kannada numeral aimed here will outsource 96% of accuracy. Keywords Pre-processing Binarization Normalization Discrete wavelet transform (DWT) Curvelet transform wrapping DCNN and handwritten Kannada numeral (HKN)
& Vishweshwrayya C. Hallur [email protected] & R. S. Hegadi [email protected] 1
Department of Computer Science, Bharathiar University, Coimbatore, Tamilnadu, India
2
School of Computational Sciences, Solapur University, Solapur, Maharashtra 413255, India
1 Introduction There are petition fields like Indian offices (bank, railway, sales tax, etc.) wielding both English and zonal dialects in filling forms. Many times those forms claim to be scanned. If the system is unable to recognize the characters then an image is apprehended resulting into no edit condition. Everyday massive count of such forms need to be processed in the corporative fields like banks, post offices, libraries, automatic reading systems to save maximum work and time. Even though enormous advances have been perpetrated in the recent years in the recognition of handwritten numerals but still they are causing a problem in motif recognition field [1]. Troupes of symbols constitute a string/word that represents information may be in isolated languages. Few writing tools are available today those are habituated to characterize the characters as physical entities and are handed over to the readers. Character identification is the topic under study from few decades. It plays paramount place in human machine interfacing. Along with this handwritten Kannada figures recognition is crucial in the applications as post office, banks and other institutions. Identification systems are classified as online or offline. The procedure of discovering letters or words that are attentive in digital image of handwritten text is known as offline handwritten recognition. Uttermost research work is prosecuted on the recognition of offline handwritten characters for the scripts like Devnagari, Bangla, Kannada and Tamil etc. [2]. The intention behind optical character recognition (OCR) is to pinp
Data Loading...
