A certain investigation on efficient character recognition system for ancient historical documents

Abstract

subsystem in the field of Document Image Analysis (DIA). OCR is used for newlineconverting the printed documents into a digital form. This digital form is used newlinefor several applications such as analyzing the nature of documents, newlineautomation of office tasks with the aim of the paperless office and converting newlineone script into another script understandable by human being. This OCR is newlinealso useful for preserving the document information for future without any newlineloss. The development of OCR has gained a lot of interest from the newlineresearchers in the field of document image analysis. Here, the degraded newlinehistorical document recognition system offers a better estimation of machine newlineeditable format. newlineIn the first part of the research, an optimum binarization technique is newlineproposed for recognizing the ancient historical documents through newlineinformation acquisition approach. The benchmark dataset namely Double newlineImage Binarization Contest (DIBCO) from pattern recognition and image newlineanalysis (PRImA) research lab is considered for analysis. The binarization newlineprocess consists of canny edge map detector followed by modified adaptive newlinethresholding to segment the foreground and the background pixels in an input newlineimage. The performance of the proposed binarization method is measured newlineusing F-measure (FM), Misclassification Penalty Metric (MPM), Negative newlineRate Metric (NRM) and Peak Signal to Noise Ratio (PSNR) and its newlinecorresponding attained values are 88.4461, 0.0708, 0.00265 and 18.420 newlinerespectively. The proposed method overcomes different degradation factors newlinelike non-uniform illumination and complex background newline