ResNet18-BLSTM-CTC-Based OCR Model for Extracting Technical Parameter Data from Digital Television Analyzer Screenshots

Henrian Robby Fakhriannur; Alva Hendi Muhammad

doi:10.34148/teknika.v15i2.1501

Authors

Henrian Robby Fakhriannur Department of Informatics, Faculty of Computer Science, Universitas Amikom Yogyakarta, Sleman, DI Yogyakarta, Indonesia
Alva Hendi Muhammad Department of Informatics, Faculty of Computer Science, Universitas Amikom Yogyakarta, Sleman, DI Yogyakarta, Indonesia

DOI:

https://doi.org/10.34148/teknika.v15i2.1501

Keywords:

Optical Character Recognition, ResNet18, Bidirectional Long Short-Term Memory, Connectionist Temporal Classification, Deep Learning

Abstract

Monitoring digital television broadcasting stations requires accurate extraction of technical measurement parameters from screenshot images generated by television analyzer devices. In current practice, parameters such as frequency, signal level, bandwidth, modulation, guard interval, code rate, Fast Fourier Transform (FFT), pilot pattern, provider name, and video standard are often read and recorded manually, which may cause transcription errors when the images contain small characters, low resolution, noise, or visually similar symbols. This study proposes an Optical Character Recognition (OCR) system based on a hybrid ResNet18-Bidirectional Long Short-Term Memory-Connectionist Temporal Classification (ResNet18-BLSTM-CTC) architecture to improve technical parameter extraction from digital television analyzer screenshots. ResNet18 is used to extract visual character features, BLSTM models sequential relationships from both directions, and CTC decodes character sequences without explicit character segmentation. The implementation includes image acquisition, Region of Interest (ROI) selection, cropping, grayscale conversion, resizing into 48 x 48 pixel character images, model training, fine-tuning using hard labels, comparative evaluation, and web-based deployment using Streamlit. After fine-tuning, the proposed model achieved 96.58% accuracy, 0.0342 Character Error Rate (CER), 0.9253 macro F1-score, and reduced errors from 80 to 55. Additional comparison with Tesseract, EasyOCR, and CRNN-CTC, as well as testing on a different analyzer dataset, shows the practical relevance of the proposed OCR workflow for digital television technical monitoring.

Downloads

Download data is not yet available.

References

[1] L. Mosbah, I. Moalla, T. M. Hamdani, B. Neji, T. Beyrouthy, and A. M. Alimi, "ADOCRNet: A Deep Learning OCR for Arabic Documents Recognition," IEEE Access, vol. 12, no. January, pp. 55620-55631, 2024, doi: 10.1109/access.2024.3379530.

[2] M. Sinthuja, C. G. Padubidri, G. S. Jayachandra, M. C. Teja, and G. S. P. Kumar, "Extraction of Text from Images Using Deep Learning," presented at the Procedia Computer Science, 2024.

[3] S. Mahadevkar, S. Patil, and K. Kotecha, "Enhancement of handwritten text recognition using AI-based hybrid approach," MethodsX, vol. 12, Jun. 2024 2024, doi: 10.1016/j.mex.2024.102654.

[4] A. A. A. Alshawi, J. Tanha, and M. A. Balafar, "An Attention-Based Convolutional Recurrent Neural Networks for Scene Text Recognition," IEEE Access, vol. 12, pp. 8123-8134, 2024, doi: 10.1109/access.2024.3352748.

[5] B. H. Nayef, S. N. H. S. Abdullah, R. Sulaiman, and A. M. Saeed, "Text Extraction with Optimal Bi-LSTM," Computers, Materials & Continua, vol. 76, no. 3, pp. 3549-3567, 2023, doi: 10.32604/cmc.2023.039528.

[6] B. Imane, A. Alae, K. Ghizlane, and M. Mrabti, "Enhancing Arabic handwritten word recognition: a CNN-BiLSTM-CTC architecture with attention mechanism and adaptive augmentation," Discover Applied Sciences, vol. 7, no. 5, 2025, doi: 10.1007/s42452-025-06952-z.

[7] A. A. Chandio, M. Asikuzzaman, M. R. Pickering, and M. Leghari, "Cursive Text Recognition in Natural Scene Images Using Deep Convolutional Recurrent Neural Network," IEEE Access, vol. 10, pp. 10062-10078, 2022, doi: 10.1109/access.2022.3144844.

[8] X. Lv, S. Gao, G. Miao, and Z. Chen, "Relay Protection Setting Sheet Detection and Recognition Approach Based on YOLOv8-CRNN-CTC," IEEE Access, vol. 14, no. January, pp. 19961-19969, 2026, doi: 10.1109/access.2026.3654663.

[9] B. V. Nguyen et al., "Water Meter Reading Based on Text Recognition Techniques and Deep Learning," IEEE Access, vol. 13, no. February, pp. 41422-41434, 2025, doi: 10.1109/access.2025.3547225.

[10] G. Suddul and J. F. L. Seguin, "A custom-built deep learning approach for text extraction from identity card images," International Journal of Informatics and Communication Technology, vol. 13, no. 1, pp. 34-41, Apr. 2024 2024, doi: 10.11591/ijict.v13i1.pp34-41.

[11] P. Kiatphaisansophon, D. Wanvarie, and N. Cooharojananone, "Efficient Text Bounding Box Identification Using Mask R-CNN: Case of Thai Documents," IEEE Access, vol. 12, pp. 49306-49328, 2024, doi: 10.1109/access.2024.3383911.

[12] A. Sarhan et al., "Egyptian car plate recognition based on YOLOv8, Easy-OCR, and CNN," Journal of Electrical Systems and Information Technology, vol. 11, no. 1, p. 32, 2024, doi: 10.1186/s43067-024-00156-y.

[13] G. Pandey, K. K. C, N. Lamichhane, and U. Subedi, "CNN based System for Automatic Number Plate Recognition," Journal of Soft Computing Paradigm, vol. 5, no. 4, pp. 347-364, Dec. 2023 2023, doi: 10.36548/jscp.2023.4.002.

[14] A. K. A. Abdulsalam, F. Saeedi, and M. A. Ambusaidi, "Robust Automatic License Plate Recognition Using Synthetic Data and Transformer-Based Deep Learning Model," IEEE Access, vol. 13, no. September, pp. 182890-182901, 2025, doi: 10.1109/access.2025.3621237.

[15] S. Fang, Z. Mao, H. Xie, Y. Wang, C. Yan, and Y. Zhang, "ABINet++: Autonomous, Bidirectional and Iterative Language Modeling for Scene Text Spotting," IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 45, no. 6, pp. 7123-7141, Jun. 2023 2023, doi: 10.1109/tpami.2022.3223908.

[16] H. Zhang, Q. Yao, J. T. Kwok, and X. Bai, "Searching a High Performance Feature Extractor for Text Recognition Network," IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 45, no. 5, pp. 6231-6246, May 2023 2023, doi: 10.1109/tpami.2022.3205748.

[17] C. Xue, J. Huang, W. Zhang, S. Lu, C. Wang, and S. Bai, "Image-to-Character-to-Word Transformers for Accurate Scene Text Recognition," IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 45, no. 11, pp. 12908-12921, Nov. 2023 2023, doi: 10.1109/tpami.2022.3230962.

[18] S. A. Raza, M. S. Farooq, U. Farooq, H. Karamti, T. Khurshaid, and I. Ashraf, "A Convolutional Neural Network Based Optical Character Recognition for Purely Handwritten Characters and Digits," Computers, Materials & Continua, vol. 84, no. 2, pp. 3149-3173, 2025, doi: 10.32604/cmc.2025.063255.

[19] H. M. R. U. Rehman, S. A. Haider, H. Faisal, K. Y. Yoo, M. Z. Jhandir, and G. S. Choi, "A Novel Framework for Saraiki Script Recognition Using Advanced Machine Learning Models (YOLOv8 and CNN)," IEEE Access, vol. 13, no. April, pp. 56843-56860, 2025, doi: 10.1109/access.2025.3551747.

[20] F. Yasir and M. Kazmi, "Acceleration of Urdu Optical Character Recognition on Zynq UltraScale+ MPSoC Using Deep Convolutional Neural Network," IEEE Access, vol. 13, no. August, pp. 135538-135557, 2025, doi: 10.1109/access.2025.3593294.

[21] C. Mai, P. Penava, and R. Buettner, "Oracle Bone Inscription Character Recognition Based on a Novel Convolutional Neural Network Architecture," IEEE Access, vol. 12, no. December, pp. 197021-197034, 2024, doi: 10.1109/access.2024.3521319.

[22] Prathwini, A. P. Rodrigues, P. Vijaya, and R. Fernandes, "Tulu Language Text Recognition and Translation," IEEE Access, vol. 12, no. January, pp. 12734-12744, 2024, doi: 10.1109/access.2024.3355470.

[23] N. Khan et al., "Systematic Literature Review of Machine Learning Models and Applications for Text Recognition," IEEE Access, vol. 13, no. August, pp. 177647-177670, 2025, doi: 10.1109/access.2025.3618109.

[24] E. Vidal, A. H. Toselli, A. Ríos-Vila, and J. Calvo-Zaragoza, "End-to-End page-Level assessment of handwritten text recognition," Pattern Recognition, vol. 142, Oct. 2023 2023, doi: 10.1016/j.patcog.2023.109695.

[25] X. Q. Liu, X. Y. Ding, X. Luo, and X. S. Xu, "ProtoUDA: Prototype-Based Unsupervised Adaptation for Cross-Domain Text Recognition," IEEE Transactions on Knowledge and Data Engineering, vol. 36, no. 12, pp. 9096-9108, 2024, doi: 10.1109/tkde.2023.3344761.

[26] H. Ren, J. Deng, X. Xie, X. Ma, and Y. Wang, "FedBoosting: Federated learning with gradient protected boosting for text recognition," Neurocomputing, vol. 569, Feb. 2024 2024, doi: 10.1016/j.neucom.2023.127126.

[27] Y. Zhang and G. Li, "Improving Handwritten Mathematical Expression Recognition via Integrating Convolutional Neural Network With Transformer and Diffusion-Based Data Augmentation," IEEE Access, vol. 12, no. May, pp. 67945-67956, 2024, doi: 10.1109/access.2024.3399919.

[28] R. Malhotra and M. T. Addis, "Handwritten Amharic Word Recognition With Additive Attention Mechanism," IEEE Access, vol. 12, no. August, pp. 114645-114657, 2024, doi: 10.1109/access.2024.3444897.

[29] T. Guan, W. Shen, X. Yang, Q. Feng, Z. Jiang, and X. Yang, "Self-supervised Character-to-Character Distillation for Text Recognition," presented at the Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), 2023. [Online]. Available: https://github.com/TongkunGuan/CCD.