Template Matching Approach for Printed Kannada Numeral Recognition

In this paper a simple template matching method based on correlation coefficient is proposed to recognize the printed Kannada numerals. Here the scanned printed Kannada numeral documents are preprocessed and each numeral is extracted by first performing the line segmentation and then segmenting each numeral from the segmented line. Each segmented object is resized to a predefined size. The correlation coefficient is computed between the image under consideration and stored numeral image data. A high value of correlation coefficient will indicate the successful match. The proposed algorithm is invariant to size of the numerals since segmented objects are resized in the preprocessing stage. Experimentation is carried out on 30 different fonts of Kannada numerals generated from Nudi 4.0 software. The results are encouraging.