Ensemble Classifier System for Offline Ancient Tamil Character Recognition
Merline Magrina M, San thi M · International Journal of Electronics and Communication Engineering · 2019
Recognizing ancient Tamil characters enable archaeologists to reveal historical events in Chola period in 12 th century with huge efforts of the archeological experts.The future research in the field of archeology will have negative impact due to inefficiency of the manual procedure.Optical Character Recognition (OCR) functionality is used to recognize ancient Tamil Inscription in stones.OCR module of application i.e.Ensemble Learning is mainly focused in this paper.The collected inscriptions are preprocessed and segmented.During preprocessing the color images are converted to gray images and they are enhanced.The resultant images are applied for noise removal process using various filters.The output is applied for post processing steps to remove further noises includes opening, closing, erosion and dilation.Segmentation process is performed using bounding box and the features like orientation points & magnitude points are detected using SIFT algorithm.Characters are classified using Ensemble learning classifier and the classified characters are matched using Unicode.The exact Modern Tamil Character is mapped and their Recognition rate & Segmentation Rates are calculated and compared.