Dampak Prapemrosesan Citra dan Teks Terhadap Tingkat Akurasi OCR dan Skor ROUGE

Fajar Hamdani; Esa Fauzi

doi:10.58602/jaiti.v4i3.315

Fajar Hamdani ^{(Corresponding Author)} Universitas Widyatama Bandung
Esa Fauzi Universitas Widyatama Bandung

DOI: https://doi.org/10.58602/jaiti.v4i3.315

Keywords: Prapemrosesan Citra, OCR, PaddleOCR, NLP, T5

Abstract

Dokumen digital berformat PDF yang memuat tulisan tangan atau hasil pindaian beresolusi rendah sering kali sulit untuk diekstraksi informasinya. Penelitian ini mengkaji dampak prapemrosesan citra dan teks dalam meningkatkan akurasi proses ekstraksi menggunakan teknologi Optical Character Recognition (OCR) serta peringkasan otomatis menggunakan Natural Language Processing (NLP). Fokus penelitian tidak diarahkan pada perancangan model OCR atau NLP baru, melainkan pada signifikansi tahapan prapemrosesan terhadap kualitas teks masukan pada sistem. Tahapan prapemrosesan citra yang diterapkan bersifat adaptif, meliputi pemotongan area teks, deskew, resize bikubik, konversi grayscale, denoising, Contrast Limited Adaptive Histogram Equalization (CLAHE), Gaussian blur, adaptive threshold, dan morfologi opening. Teks hasil OCR kemudian dibersihkan dan diperbaiki secara adaptif sebelum diringkas menggunakan model T5 bahasa Indonesia. Hasil pengujian terhadap 23 sampel dokumen menunjukkan bahwa intervensi prapemrosesan berhasil meningkatkan akurasi pengenalan karakter dari 65,31% menjadi 75,10%. Selain itu, pada tahap evaluasi peringkasan teks juga terjadi peningkatan skor ROUGE-1 dari 21,89% menjadi 32,24%, ROUGE-2 dari 8,69% menjadi 16,16%, dan ROUGE-L dari 16,93% menjadi 26,43%. Hasil pengujian ini membuktikan bahwa tahapan prapemrosesan citra dan teks sangat penting dalam memperbaiki kinerja pengenalan karakter dan peringkasan dokumen

Downloads

Download data is not yet available.

References

Agustian, I. A. (2025). Implementation of the T5-small model for abstractive text summarization in SEO meta description generation. Universitas Nusa Putra.

Almegbil, L., & al., et. (2026). Raqim: A hybrid framework for correcting Arabic OCR errors by integrating dictionary-based and generative techniques. Procedia Computer Science, 275, 493–502.

Arifin, M. N., Mansyur, M. U., & dkk. (2025). Enhanced OCR recognition for Madurese text documents: A genetic algorithm approach with Tesseract 5.5. JUITA, 13(2).

Baek, Y., Lee, B., Han, D., Yun, S., & Lee, H. (2019). What is wrong with scene text recognition model comparisons? Dataset and model analysis. Proc. IEEE/CVF Int. Conf. Computer Vision, 4715–4723.

Bhaskoro, S. B., Pratama, R. A., & Darmawan, S. A. H. (2025). Deteksi dan interpretasi tulisan tangan bahasa Indonesia melalui pemrosesan citra dan optical character recognition (OCR). JTRM (Jurnal Teknologi Dan Rekayasa Manufaktur), 7(1), 48–64.

Biswas, S., & Das, A. K. (2020). Effects of morphological operations on optical character recognition for degraded text images. Journal of Visual Communication and Image Representation, 68, 102769.

Cahyawijaya, S., & al., et. (2021). IndoNLG: Benchmark and resources for evaluating Indonesian natural language generation. Proc. 2021 Conf. Empirical Methods in Natural Language Processing, 8875–8898.

Dey, R., Singh, D., Balabantaray, R. C., Karuppiah, M., Mohanty, S., & Samanta, D. (2022). Approach for preprocessing in offline optical character recognition (OCR). Proc. Interdisciplinary Research in Technology and Management (IRTM), 1–6.

Ingle, R. R., Fujii, Y., Deselaers, T., Baccash, J., & Popat, A. C. (2019). A scalable handwritten text recognition system. Proc. Int. Conf. Document Analysis and Recognition (ICDAR), 17–24.

Jiao, L., & Zhang, Y. (2019). Attention-based end-to-end neural network for handwritten character recognition. IEEE Access, 7, 9123–9130.

Koto, F., Lau, J. H., & Baldwin, T. (2020). Liputan6: A large-scale Indonesian dataset for text summarization. Proc. AACL-IJCNLP, 5982–5998.

Memon, J., Sami, M., Khan, R. A., & Uddin, M. (2020). Handwritten optical character recognition (OCR): A comprehensive systematic literature review (SLR). IEEE Access, 8, 142642–142668.

Mursari, L. R., & Wibowo, A. (2021). The effectiveness of image preprocessing on digital handwritten scripts recognition with the implementation of OCR Tesseract. Computer Engineering and Applications Journal, 10(3), 177–186.

Nugroho, A. S., & Hidayatullah, A. F. (2019). Analisis kinerja OCR Tesseract pada pengenalan karakter dokumen berbahasa Indonesia. Jurnal Sistem Informasi, 15(1), 12–20.

Purnama, I. N., & Utami, N. N. W. (2023). Implementasi peringkas dokumen berbahasa Indonesia menggunakan metode text to text transfer transformer (T5). Jurnal Teknologi Informasi Dan Komputer, 9(4), 374–383.

Raffel, C., & al., et. (2020). Exploring the limits of transfer learning with a unified text-to-text transformer. Journal of Machine Learning Research, 21(140), 1–67.

See, A., Liu, P. J., & Manning, C. D. (2017). Get to the point: Summarization with pointer-generator networks. Proc. 55th Annual Meeting of the Association for Computational Linguistics, 1073–1083.

Setiawan, A., & Purnamasari, D. (2021). Text summarization pada artikel berita berbahasa Indonesia menggunakan metode transformer. Jurnal RESTI (Rekayasa Sistem Dan Teknologi Informasi), 5(3), 450–457.

Shi, B., Bai, X., & Yao, C. (2017). An end-to-end trainable neural network for image-based sequence recognition and its application to scene text recognition. IEEE Transactions on Pattern Analysis and Machine Intelligence, 39(11), 2298–2304.

Singh, A., & Gupta, A. (2023). A review on preprocessing techniques for handwritten text recognition. Multimedia Tools and Applications, 82(10), 15555–15578.

Syafiq, M. A., & Adiwijaya, A. (2023). Kinerja model bahasa berbasis transformer untuk peringkasan teks abstraktif pada dokumen bahasa Indonesia. Jurnal Media Informatika Budidarma, 7(1), 210–218.

Talirongan, H., & al., et. (2025). Advanced image preprocessing strategies for mitigating challenges in optical character recognition of diverse textual images. International Journal of Environmental Sciences, 11(25s), 421–429.

Vaswani, A., & al., et. (2017). Attention is all you need. Advances in Neural Information Processing Systems, 30.

Vo, Q. N., Kim, S. H., Yang, H. J., & Lee, G. (2018). Binarization of degraded document images based on hierarchical deep supervised network. Pattern Recognition, 74, 568–586.

Wibowo, A. T., & Ardiansyah, F. (2020). Penerapan metode erosi dan dilasi pada citra digital untuk memperbaiki kualitas hasil OCR. Jurnal Informatika Dan Komputasi, 14(2), 112–120.

Wu, S., & Chang, F. (2025). A dual-engine fusion optical character recognition method for fast identification and key information extraction of drug labels. Alexandria Engineering Journal, 128, 1027–1036.

Zhang, Y., Liu, X., & Wang, Z. (2024). Advanced morphological image processing techniques for degraded document recognition. Pattern Recognition Letters, 178, 112–119.