Binarization of Malayalam Palm Leaf Images using VGG Net and Integrated Otsu

Binarization involves converting gray-scale images into binary files. The binarization of palm leaves is an essential stage in digitalization since it improves the readability, accessibility, and preservation of these priceless historical records. It allows for enhanced study, investigation, and worldwide distribution of the rich cultural and historical material contained within them. We experimented using the U-Net architecture model, which uses VGG as encoder and FCN decoder, also finetuning the thresholding value, making an integrated Otsu implementation that provides the best-binarized image of Peak signal-to-noise ratio of 21.11 and Mean average precision of 0.99.