An End-to-End Balinese Lontar OCR Framework Using Bayesian-Optimized Multiscale Retinex and MobileNetV3

Authors

  • Putu Ayu Febyanti Institut Bisnis dan Teknologi Indonesia
  • I Putu Agus Eka Darma Udayana Institut Bisnis dan Teknologi Indonesia
  • Aniek Suryanti Kusuma Institut Bisnis dan Teknologi Indonesia

DOI:

10.33395/sinkron.v10i3.16271

Keywords:

Adaptive preprocessing, Balinese script recognition, Bayesian optimization, MobileNetV3, Multiscale Retinex, optical character recognition, palm-leaf manuscript OCR.

Abstract

The preservation of Balinese lontar manuscripts has become increasingly important due to their cultural, historical, and religious significance, while physical degradation such as uneven illumination, faded ink, texture interference, and manuscript aging continues to reduce readability and complicate digital preservation efforts. This study proposes an end-to-end Optical Character Recognition (OCR) framework for degraded Balinese lontar manuscripts by integrating Bayesian-optimized image enhancement, adaptive preprocessing, morphology-based segmentation, domain-specific augmentation, and lightweight deep learning recognition using MobileNetV3. The proposed enhancement pipeline combines Multiscale Retinex, adaptive gamma correction, edge-preserving filtering, and hybrid binarization to improve character visibility under degraded manuscript conditions. Bayesian Optimization with Optuna and Tree-structured Parzen Estimator (TPE) was employed to automatically optimize enhancement parameters according to manuscript quality characteristics. Experimental results demonstrated substantial improvements in manuscript image quality, where Laplacian Variance increased from 306.7596 to 6685.7641, RMS Contrast improved from 28.976 to 83.9085, Michelson Contrast increased from 0.8238 to 1.0, and Ink Ratio Score improved from 0.6096 to 0.9847. The MobileNetV3-based OCR recognition model achieved a test accuracy of 80.52% and a best validation accuracy of 83.78% across 102 Balinese script classes. The proposed framework demonstrates that adaptive enhancement optimization combined with lightweight OCR recognition can provide robust and computationally efficient recognition performance for degraded historical manuscripts while supporting scalable digital preservation and mobile-oriented cultural heritage applications.

GS Cited Analysis

Downloads

Download data is not yet available.

References

Anh, T. P., & Nhan, V. N. (2022). A Novel Light-Weight DCNN Model for Classifying Plant Diseases on Internet of Things Edge Devices. Mendel, 28(2), 41–48. https://doi.org/10.13164/mendel.2022.2.041

Aung, T. (2025). CNN-Based Identification of Pathogens of Concern in Shrimp. Animals, 15(21), 3194. https://doi.org/10.3390/ani15213194

Bindhu, A., & Maheswari, O. U. (2020). Color Corrected Single Scale Retinex Based Haze Removal and Color Correction for Underwater Images. Color Research & Application, 45(6), 1084–1093. https://doi.org/10.1002/col.22568

Chi, J., On, C. K., Zhang, H., & Chai, S. S. (2023). A Review of Deep Convolutional Neural Networks in Mobile Face Recognition. International Journal of Interactive Mobile Technologies (Ijim), 17(23), 4–19. https://doi.org/10.3991/ijim.v17i23.40867

Dai, Y., Dai, Y., Xuan, S., Luo, S., Zuo, J., & Liu, X. (2025). A Tool Wear Prediction Method Based on Improved MobileNetV3. Engineering Research Express, 7(4), 45251. https://doi.org/10.1088/2631-8695/ae1715

Faizullah, S., Ayub, M. S., Hussain, S., & Khan, M. A. (2023). A Survey of OCR in Arabic Language: Applications, Techniques, and Challenges. Applied Sciences, 13(7), 4584. https://doi.org/10.3390/app13074584

Huang, F. (2020). Parallelization Implementation of the Multi‐scale Retinex Image‐enhancement Algorithm Based on a many Integrated Core Platform. Concurrency and Computation Practice and Experience, 32(22). https://doi.org/10.1002/cpe.5832

Ibrahim, M., Wagdy, M., Alharithi, F. S., Qahtani, A. M., Elkilani, W. S., & Zarif, S. (2022). An Efficient Method for Document Correction Based on Checkerboard Calibration Pattern. Applied Sciences, 12(18), 9014. https://doi.org/10.3390/app12189014

Indrawan, G., Asroni, A., Dewi, L. J. E., Gunadi, I. G. A., & Paramarta, I. K. (2022). Balinese Script Recognition Using Tesseract Mobile Framework. Lontar Komputer Jurnal Ilmiah Teknologi Informasi, 13(3), 160. https://doi.org/10.24843/lkjiti.2022.v13.i03.p03

Jia, L., Wang, Y., Zang, Y., Li, Q., Leng, H., Xiao, Z., Long, W., & Jiang, L. (2022). MobileNetV3 With CBAM for Bamboo Stick Counting. Ieee Access, 10, 53963–53971. https://doi.org/10.1109/access.2022.3175818

Kang, M., Park, R., & Park, H. (2021). Efficient Spatio-Temporal Modeling Methods for Real-Time Violence Recognition. Ieee Access, 9, 76270–76285. https://doi.org/10.1109/access.2021.3083273

Kesiman, M. W. A., & Pradnyana, G. A. (2021). A Scheme Towards Automatic Word Indexation System for Balinese Palm Leaf Manuscripts. Journal of Ict Research and Applications, 15(2), 105–119. https://doi.org/10.5614/itbj.ict.res.appl.2021.15.2.1

Kiran, V. K., Dash, S., & Parida, P. (2024). Vehicle Detection in Varied Weather Conditions Using Enhanced Deep YOLO With Complex Wavelet. Engineering Research Express, 6(2), 25224. https://doi.org/10.1088/2631-8695/ad507d

Koli, S. (2025). Sustainable Edge AI for Precision Agriculture: A Lightweight CNN Model for Aloe Vera Leaf Disease Diagnosis. Expert Systems, 42(12). https://doi.org/10.1111/exsy.70154

Krithiga, R., Varsini, S. R., Joshua, R., & Kumar, C. U. O. (2025). Ancient Character Recognition: A Comprehensive Review. Ieee Access, 13, 88847–88857. https://doi.org/10.1109/access.2023.3341352

Labib, M., Amin, K., & Ibrahim, M. (2021). Warped Document Image Correction Based on Checkboard Pattern and Geometric Transformation. Ijci International Journal of Computers and Information, 0(0), 0. https://doi.org/10.21608/ijci.2021.53176.1036

Li, L., Yang, H., & Wang, B. (2023). Cerasus Humilis Cultivars Identification With Small-Sample and Unbalanced Dataset Based on Efficient Net-B0+ranger Networks. Inmateh Agricultural Engineering, 225–236. https://doi.org/10.35633/inmateh-69-21

Li, S., Zhao, S., Gao, P., Ge, S., Huang, R., & Shu, L. (2025). HP-MobileNetV3: A Lightweight Neural Network for Non-Intrusive Load Monitoring. Engineering Research Express, 7(1), 15223. https://doi.org/10.1088/2631-8695/ada66a

Li, W., Lin, H., Yan, Y., & Feng, Z. (2025). Micro-Expression Recognition Based on the Fusion of Facial Landmark Points and Optical Flow Features. Journal of Electronic Imaging, 34(01). https://doi.org/10.1117/1.jei.34.1.013028

Lisani, J.-L., Petro, A. B., Sbert, C., Álvarez-Ellacuría, A., Catalán, I. A., & Palmer, M. (2022). Analysis of Underwater Image Processing Methods for Annotation in Deep Learning Based Fish Detection. Ieee Access, 10, 130359–130372. https://doi.org/10.1109/access.2022.3227026

Liu, J., Mao, Q., Yan, L., Zeng, L., Zou, Y., Li, J., Li, Y., Luo, Z., Huang, Z., & Xiao, W. (2024). A Novel Enhancing Method for Terahertz Imaging of Integrated Circuits Flaw Detection. Physica Scripta, 99(11), 115116. https://doi.org/10.1088/1402-4896/ad8400

Lombardi, F., & Marinai, S. (2020). Deep Learning for Historical Document Analysis and Recognition—A Survey. Journal of Imaging, 6(10), 110. https://doi.org/10.3390/jimaging6100110

Mahadevkar, S., Patil, S., Kotecha, K., & Abraham, A. (2024). A Comparison of Deep Transfer Learning Backbone Architecture Techniques for Printed Text Detection of Different Font Styles From Unstructured Documents. Peerj Computer Science, 10, e1769. https://doi.org/10.7717/peerj-cs.1769

Maheswari, S. U., Maheswari, P. U., & Aakaash, G. R. S. (2024). An intelligent character segmentation system coupled with deep learning based recognition for the digitization of ancient Tamil palm leaf manuscripts. Heritage Science, 12(1), 342. https://doi.org/10.1186/s40494-024-01438-4

Malik, S. A., Sajid, A., Ahmad, A., Almogren, A., Hayat, B., Awais, M., & Kim, K. H. (2020). An Efficient Skewed Line Segmentation Technique for Cursive Script OCR. Scientific Programming, 2020, 1–12. https://doi.org/10.1155/2020/8866041

Manuaba, P., & Indah, K. A. T. (2021). The Object Detection System of Balinese Script on Traditional Balinese Manuscript With Findcontours Method. Matrix Jurnal Manajemen Teknologi Dan Informatika, 11(3), 177–184. https://doi.org/10.31940/matrix.v11i3.177-184

Mao, Q., Yunlong, Z., & liu, jingbo. (2022). Terahertz Image Enhancing Based on the Physical Model and Multiscale Retinex Algorithm. Applied Optics, 61(28), 8382. https://doi.org/10.1364/ao.470769

Peng, H., Xu, H., Shen, G., Liu, H., Guan, X., & Li, M. (2024). A Lightweight Crop Pest Classification Method Based on Improved MobileNet-V2 Model. Agronomy, 14(6), 1334. https://doi.org/10.3390/agronomy14061334

Prasetiadi, A., Saputra, J., Kresna, I., & Ramadhanti, I. (2023). YOLOv5 and U-Net-Based Character Detection for Nusantara Script. Jurnal Online Informatika, 8(2), 232–241. https://doi.org/10.15575/join.v8i2.1180

Pratama, A. A., Sulistiyo, M. D., & Ihsan, A. F. (2023). Balinese Script Handwriting Recognition Using Faster R-CNN. Jurnal Resti (Rekayasa Sistem Dan Teknologi Informasi), 7(6), 1268–1275. https://doi.org/10.29207/resti.v7i6.5176

Ruwanmini, S., Dias, K., Niluckshini, C., & Nandasara, T. (2023). Sinhala Inscription Character Recognition Model Using Deep Learning Technologies. International Journal on Advances in Ict for Emerging Regions (Icter), 16(1), 1–11. https://doi.org/10.4038/icter.v16i1.7239

Siahaan, D., Sutramiani, N. P., Suciati, N., Duija, I. N., & Darma, I. W. A. S. (2022a). DeepLontar Dataset for Handwritten Balinese Character Detection and Syllable Recognition on Lontar Manuscript. Scientific Data, 9(1). https://doi.org/10.1038/s41597-022-01867-5

Suciati, N., Sutramiani, N. P., & Siahaan, D. (2022). LONTAR_DETC: Dense and High Variance Balinese Character Detection Method in Lontar Manuscripts. Ieee Access, 10, 14600–14609. https://doi.org/10.1109/access.2022.3147069

Sudarsan, D., & Sankar, D. (2024). An Ensemble Neural Network Model For Malayalam Character Recognition From Palm Leaf Manuscripts. ACM Trans. Asian Low-Resour. Lang. Inf. Process. https://doi.org/10.1145/3686311

Sun, Y., Zhao, Z., Jiang, D., Tong, X., Tao, B., Jiang, G., Kong, J., Yun, J., Liu, Y., Liu, X., Zhao, G., & Fang, Z. (2022). Low-Illumination Image Enhancement Algorithm Based on Improved Multi-Scale Retinex and ABC Algorithm Optimization. Frontiers in Bioengineering and Biotechnology, 10. https://doi.org/10.3389/fbioe.2022.865820

Supriyanto, S., Harika, M., Ramadiani, M. S., & Ramdania, D. R. (2020). Multiscale Retinex Application to Analyze Face Recognition. Jurnal Online Informatika, 217–226. https://doi.org/10.15575/join.v5i2.668

Sutramiani, N. P., Darma, I. W. A. S., & Arsa, D. M. S. (2022). Handwritten Balinese Script Recognition on Palm Leaf Manuscript Using Projection Profile and K-Nearest Neighbor. Jurnal Ilmiah Merpati (Menara Penelitian Akademika Teknologi Informasi), 10(3), 133. https://doi.org/10.24843/jim.2022.v10.i03.p02

Wang, Q., Zhang, N., Jiang, K., Ma, C., Zhou, Z., & Dai, C. (2021). Tunnel Lining Crack Recognition Based on Improved Multiscale Retinex and Sobel Edge Detection. Mathematical Problems in Engineering, 2021, 1–9. https://doi.org/10.1155/2021/9969464

Xiao, Y., Jiang, A., Ye, J., & Wang, M. (2020). Making of Night Vision: Object Detection Under Low-Illumination. Ieee Access, 8, 123075–123086. https://doi.org/10.1109/access.2020.3007610

Yan, C. (2025). TinyML ‐Enhanced Cloud‐Edge Collaborative Framework for Real‐Time Sport Action Recognition. Internet Technology Letters, 8(5). https://doi.org/10.1002/itl2.70100

Yoon, K., Lee, S., Park, J., & Kim, K. G. (2025). CNN-Based Automatic Tablet Classification Using a Vibration-Controlled Bowl Feeder With Spiral Torque Optimization. Sensors, 25(14), 4248. https://doi.org/10.3390/s25144248

Yuadi, I., Nisa’, K., Nazikhah, N. U., Halim, Y. A., Asyhari, A. T., & Hu, C.-C. (2025). A Benchmark Study of Classical and U-Net ResNet34 Methods for Binarization of Balinese Palm Leaf Manuscripts. Heritage, 8(8). https://doi.org/10.3390/heritage8080337

Zhang, A., Li, Y., & Wang, S. (2024). 2ddsru-MobileNet: An End-to-End Cloud-Noise-Robust Lightweight Convolution Neural Network. Journal of Applied Remote Sensing, 18(02). https://doi.org/10.1117/1.jrs.18.024511

Zhang, E., Guo, L., Guo, J., Yan, S., Li, X., & Kong, L. (2023). A Low-Brightness Image Enhancement Algorithm Based on Multi-Scale Fusion. Applied Sciences, 13(18), 10230. https://doi.org/10.3390/app131810230

Zhang, J., Yang, X., Fu, X., Wang, B., & Li, H. (2025). LDL-MobileNetV3S: An Enhanced Lightweight MobileNetV3-small Model for Potato Leaf Disease Diagnosis Through Multi-Module Fusion. Frontiers in Plant Science, 16. https://doi.org/10.3389/fpls.2025.1656731

Zhang, M., Zhang, Y., Jiang, Z., Lv, X., & Guo, C. (2021). Low-Illumination Image Enhancement in the Space Environment Based on the DC-WGAN Algorithm. Sensors, 21(1), 286. https://doi.org/10.3390/s21010286

Downloads


Crossmark Updates

How to Cite

Febyanti, P. A., I Putu Agus Eka Darma Udayana, & Aniek Suryanti Kusuma. (2026). An End-to-End Balinese Lontar OCR Framework Using Bayesian-Optimized Multiscale Retinex and MobileNetV3. Sinkron : Jurnal Dan Penelitian Teknik Informatika, 10(3), 1295-1310. https://doi.org/10.33395/sinkron.v10i3.16271

Issue

Vol. 10 No. 3 (2026): Article Research July 2026

Section

Articles

License

Copyright (c) 2026 Putu Ayu Febyanti, I Putu Agus Eka Darma Udayana, Aniek Suryanti Kusuma

Creative Commons License

This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.

Crossref
Scopus
Google Scholar
Europe PMC