A binarization method for Odia palm leaf manuscripts by using the Odia language accelerator model
Aloka Natha, Bichitrananda Behera, Nishanta Ranjan Nanda · 2025
A binarization technique is suggested for improving the images of palm leaf manuscripts. It can efficiently eliminate unnecessary lines from the photos of palm leaf manuscripts and perform image binarization. Using machine learning, an enhanced background elimination technique is first introduced to address the issue of various backdrop colors in a palm leaf. Following that, the image is binarized using the AI-based large language model (LLM) technique. However, the result photographs produced by the above include a lot of noise, therefore noise is eliminated using a noise reduction technique. Lastly, a connected domain-based OCR technique is explained for removing unnecessary lines with impressions and Odia letters on palm leaves. Experimental findings from Python programming demonstrate how successful the suggested approach is.