Performance Comparison of Gurmukhi Script: k-NN Classifier with DCT and Gabor Filter

Sapna Dhiman, Dr. Gurpreet Singh Lehal


This paper presents a comparative performance analysis for Gurmukhi OCR at word level. To evaluate the performance k¬-NN classifier has been used. Before the classification, Features have been extracted from word images. For feature extraction, word images have been scanned and these images are machine printed images.Here Discrete Cosine Transform (DCT) and Gabor filter has been used to extract the features. DCT provides 100 features of scanned images in zig-zag method and Gobor provides 189 features for scanned images. To train the classifier of Gurmukhi OCR, 50 different classes with 30-35 samples of each class i.e 1600 samples have been taken. 750 samples have been used to test the system. Using Gabor filter, k-NN classifier provides 92.6229%of correctness while with DCT with k-NN provides 96.9945% of accuracy.


Feature extraction, Gabor Filter, Discrete Cosine Transform (DCT), Classifier, k-NN, OCR

Full Text:




  • There are currently no refbacks.

Copyright (c) 2017 International Journal of Advanced Research in Computer Science