Offline Handwritten Character Recognition for Gujarati Language |
Author(s): |
sumit trivedi , l.d.college of engineering; arun nandurbarkar, l.d.colllege of engineering; |
Keywords: |
OCR, HCR, Gujarati Language, Feature Extraction, SVM, k-NN, Naïve Bayes |
Abstract |
This paper deals with the problem of recognizing handwritten Gujarati characters from scanned image. First, pre-processing steps are applied on input scanned image. i.e. RGB to binary conversion, removal of noise and different morphological operation to find skeleton of an image and normalization of each character image to 30x30 sizes. Second, Features are extracted from the normalized 30x30 image for each character. Here we have extracted gradient feature and also proposed a new combination of features which is combination of correlation function based feature, invariant moments and projection profiles. Third, these extracted features are supplied as an i/p to classifiers. i.e. SVM, k-NN and Naïve Bayes. The performance of different classifiers measured using 10-fold cross validation. |
Other Details |
Paper ID: NCACSETT4P057 Published in: Conference 10 : NCACSET 2017 Publication Date: 06/05/2017 Page(s): 136-139 |
Article Preview |
Download Article |
|