Classification and Recognition of Gujarati Numeral using Nearest Neighbour Approach
Hetal R.Thaker, Vivek J. Vyas, Vaishali G. Sanghvi · Zenodo (CERN European Organization for Nuclear Research) · 2016
Optical Character Recognition is an area much explored since the beginning due to its wide range of applications. This paper focuses on optical Gujarati digit recognition. To recognize Gujarati digit certain set of operations like pre-processing is carried out. As a result of feature extraction vector of 9 elements is derived representing no. of connected neighbor pixels and last element presenting total of 8 elements. Classification repository is derived by applying pre-processing and feature extraction steps for each digit to create classification repository matrix of 10 X 9 size, where row represents Gujarati digit 0 to 9. To test this model image containing isolated Gujarati digit is given as input and the result of feature extraction vector is compared with last element of each row of classification repository matrix. If total matches then digit is recognized on the basis of index of the row. In case of similar total for two digits each element of vector is compared with row of classification repository matrix. Methodology, implementation of it and results obtained are presented in this paper.