Performance Evaluation of pre-processing techniques for historical Palm Leaf Manuscript image restoration

In olden days, enormous information was noted, stored and conveyed using palm leaves as it was considered a reliable source for preserving it. Nowadays most of these leaves are deteriorated and posing a threat in the degradation of the treasured information. Digitization is the only viable solution to preserve this information. Scanning and digitizing them eases creating it detectible and legible for the readers. This process initiated with the main stage of image restoration which will aid for character recognition as it is a difficult job owing to the diversity of different handwritings on the palm leaves. In this work, Cellular Automata (CA) based optimal edge detection, Otsu's algorithm and Hybrid Isotropic Diffusion with the Perona-Malik (HIDPM) Model have been employed for image extraction from a palm-leaf manuscript and performances were compared with the standard edge detectors. The digitization of the palm-leaf manuscript will preserve their respective zonal activities in the field of astronomy, astrology, law, medicines, etc. This digitization will be helpful for the archaeologists or research learners for the retrieval of information very easily thereby aiding in the preservation of original manuscripts that are subjected to degradation.