Effective Thinning Algorithm for Recognition of Hand Written Devnagri Compound Characters Using Neural Network

Abstract

Handwriting always works as a primary tool in the world of communication. Various techniques have been and are developed in order to document the handwritten text. There are certain exceptional techniques that describe the environment of handwritten scripts and further convert it into electronic data by implementing various algorithms. Devnagri is widely used for most popular and commonly used languages like Marathi and Hindi We have proposed a neural network based hand written Devnagri compound character recognition system using a novel rule-based thinning algorithm that improves the overall performance of the system. The distinctive feature of our thinning algorithm is that it is a rotation invariant method which thins characters to their central lines, helping in preserving shape of the character. The system applies different rules concurrently to each pixel in the image which results into symmetrical thinning and improvement in the overall speed of the system. Results obtained for the system prove that the system is efficient enough to preserve the topology of the compound characters written in Devnagari that further helps in improving the accuracy of the recognition system.

Automatic Numeral Recognition System Using Local Statistical and Geometrical Features

“…The accuracy of license plate recognition was about 68.57%, while the accuracy of character recognition on the license plate was approximately 92.72%. Many research domains have employed thinning as a preprocessing step, including numeral recognition [15], character recognition [19], image coding and biometrics [20], and fingerprint identification [19]. Where the result of thinning is utilized in determining the spin direction of the fingerprint, as in [20,21].…”