Ancient Tamil Character Recognition from Epigraphical Inscriptions using Image Processing Techniques

Abstract

Ancient Tamil characters are recognized and the archeologist are enabled to reveal the 12th century Historical events during Chola period. Due to some inefficiency in the procedure, the research in the field of archeology had a negative impact. By using OCR methodology the ancient Tamil characters in stones are recognized. Through morphological operations the input image follows the steps like pre-processing and subjected to segmentation. The color images are converted to gray image, enhanced and, salt and pepper noise removed. The bounding boxes are used to segment the dilated image through which the features like region properties, corner points, magnitude, and orientation key-points are extracted. Ensemble learning and KNN are used to classify the characters and by using Unicode the classified characters are matched. The segmented rate and Recognition rate are calculated and compared by mapping the exact modern Tamil characters.