Handwritten Marathi Character Recognition using Neural Networks
Single-stage, Multistage and Hybrid Classification techniques
Versandkostenfrei!
Versandfertig in 6-10 Tagen
44,99 €
inkl. MwSt.
PAYBACK Punkte
22 °P sammeln!
This book presents a comprehensive research on handwritten Marathi character recognition using Neural networks. It includes extensive study on how to use Neural networks efficiently and achieve high recognition results. Various issues related to Neural networks like selection of the architecture, learning algorithm, number of features, number of hidden neurons etc. are discussed in detail. It includes investigations in pre-processing and features extraction techniques. It explores the combination of Neural networks with other classifiers in various ways to handle a large amount of training dat...
This book presents a comprehensive research on handwritten Marathi character recognition using Neural networks. It includes extensive study on how to use Neural networks efficiently and achieve high recognition results. Various issues related to Neural networks like selection of the architecture, learning algorithm, number of features, number of hidden neurons etc. are discussed in detail. It includes investigations in pre-processing and features extraction techniques. It explores the combination of Neural networks with other classifiers in various ways to handle a large amount of training data successfully. It presents various classifier combinations, both in horizontal and vertical manner that increases the recognition rate without much computational complexities. This book serves an excellent baseline for those interested in carrying out research in handwritten character recognition. The algorithms discussed here can be applied to recognize characters in other scripts with respective database and structural classification parameters. The structural classification, compound character recognition and similar character recognition are the major contributions of this research.
Andere Kunden interessierten sich für
