PLM-Res-U-Net: A light weight binarization model for enhancement of multi-textured palm leaf manuscript images

• Published in: Digital Applications in Archaeology and Cultural Heritage 2024-09, Vol.34, p.e00360, Article e00360
• Main Authors: Rani, N. Shobha, Akhilesh, T.M., Bipin Nair, B.J., Koushik, K.S., Smith, Elisa Barney
• Format: Article
• Language: English
• Subjects: Deep semantic segmentation; Image enhancement; Machine Learning; Maskininlärning; Noise removal; Palm leaf manuscripts; Textured background
• ISSN: 2212-0548; 2212-0548
• DOI: 10.1016/j.daach.2024.e00360

This paper proposes a deep semantic binarization model, PLM-Res-U-Net, for enhancing palm-leaf manuscripts. PLM-Res-U-Net is a lightweight model comprising encoding and decoding blocks with skip connections. The model enhances the palm leaf manuscript by efficiently retaining the text strokes by removing the degradations such as uneven illumination, aging marks, brittleness, and background discolorations. Two datasets of palm leaf manuscript collections with multiple degradation patterns and diverse textured backgrounds are used for experimentation. PLM-Res-U-Net is trained from scratch with 50 epochs with a learning rate of1e−8 with three sampling strategies. The performance of state-of-the-art deep learning models ResUnet, Pspnet, U-Net++, and Segnet are also evaluated along with two diverse benchmark datasets. Analysis shows that results obtained by the proposed PLM-Res-U-Net prove generalizability and computational efficacy with a dice score of 0.986. Additionally, PLM-Res-U-Net successfully preserves the edge strokes of the text compared with state-of-the-art models. •The work addresses the challenges of enhancing multi-textured and degraded palm-leaf manuscripts.•A lightweight deep semantic binarization model is introduced with encoding, decoding blocks, and skip connections.•Multi-textured datasets are employed to evaluate the model's performance and assess generalizability.•The model is trained from scratch with three sampling strategies to enhance the model's generalizability.•PLM-Res-U-Net is compared with state-of-the-art deep learning models to prove its advantage