Curvelet Transform based approach for prediction of epigraphical scripts era

Classification of epigraphical scripts into various era is one of the major challenges in the field of document image analysis and recognition. The shapes of the character set in epigraphical scripts have varied over the eras. To understand the script, it is necessary to know the corresponding era and its character set. Lines and curves are the dominating features of these character sets. Since curvelet transform is effective to handle these features, in this paper Fast Discrete Curvelet Transform (FDCT) based model is designed to predict the era of the script. Experimentation is conducted on a data set comprising of 4145 images belonging to six different eras. The recognition result of the proposed method is 85.78%. The proposed method is compared with Gabor filter and Zernike moments based approaches. The results show that the proposed method on an average has 20% to 25% accuracy over Zernike moment based and Gabor filter approaches in predicting the eras of the epigraphical scripts.