A Symbolic Representation In Historical Kannada Document

Abstract

In India, the Kannada is one of the historical and formal languages in Karnataka. The historical Kannada documents gives us information about education, legislation, culture and traditions that have been practiced. Getting such information from stone carvings and palm leaves and other sources are improves our knowledge of Kannada language. Extracting information from historical records is very challenging because of poor quality, variability, contrast and envelope of characters. In this work, the given document is pre-processed using connected component analysis. The features like LBP, Gabor filter and GLTP are extracted. The features are stored in knowledgebase by aggregating and representing as inter-value type data. Symbolic classifier is introduced for the classification purpose. Experiment was conducted on our database to verify the performance of the presented method.