Home / Current Issue / Paper 1715080
Deblurring of Text Images
Subject area: Science,Engineering and Technology · Area of research: Dip
DOI: https://doi.org/10.64388/IREV9I9-1715080
Abstract
Blur in photographed or scanned text documents is a frequent and damaging degradation that reduces character legibility and OCR accuracy. This paper addresses blind text image deblurring, where neither the blur kernel nor its parameters are known in advance. The method builds a Text-Specific Hybrid Dictionary (THD) from three categories of patch pairs—Gaussian blur–sharp, motion blur–sharp, and sharp–sharp—that together cover the range of states an iterative solver encounters as it refines its latent image estimate. A Text Property (TP) Enhancement Operator, based on anchored regression on the trained dictionary, projects each intermediate patch toward the space of sharp text at every iteration. The overall objective is minimized by alternating between latent image and blur kernel subproblems using half-quadratic splitting. Final restoration uses a non-blind deconvolution step once the kernel is reliably estimated. Experiments on both synthetic and real-world blurred text images show consistent improvements over existing methods in PSNR, SSIM, and kernel similarity, with a corresponding gain in OCR accuracy. The experimental results demonstrate that the proposed approach effectively restores blurred text images and significantly improves text readability and OCR performance compared with existing techniques
Keywords
Blind Text Image Deblurring, Hybrid Dictionary, Sparse Representation, Alternating Optimization, PSNR, SSIM, OCR Accuracy.
References
[1] D. Krishnan, T. Tay, and R. Fergus, "Blind deconvolution using a normalized sparsity measure," in Proc. IEEE CVPR, Jun. 2011, pp. 233-240.
[2] Q. Shan, J. Jia, and A. Agarwala, "High- quality motion deblurring from a single image," ACM Trans. Graph., vol. 27, no. 3, 2008.
[3] T.-H. Li and K.-S. Lii, "A joint estimation approach for two-tone image deblurring," IEEE Trans. Image Process., vol. 11, no. 8, pp. 847- 858, Aug. 2002.
[4] J. Pan, Z. Hu, Z. Su, and M.-H. Yang, "Deblurring text images via L0-regularized Method PSNR (dB) SSIM MSE OCR Acc. (%) KS Blurred Input 21.3 0.71 0.0048 61.2 0.63 Krishnan
[1] 26.8 0.83 0.0021 75.4 0.80 Pan
[4] 28.4 0.86 0.0015 81.7 0.85 Cao
[10] 29.7 0.88 0.0011 85.3 0.88 Proposed (THD) 31.2 0.91 0.0008 90.6 0.92 intensity and gradient prior," in Proc. IEEE CVPR, Jun. 2014, pp. 2901-2908.
[5] H. Cho, J. Wang, and S. Lee, "Text image deblurring using text-specific properties," in Proc. ECCV, Oct. 2012, pp. 524-537.
[6] H. Lee, C. Jung, and C. Kim, "Blind deblurring of text images using a text-specific hybrid dictionary," IEEE Trans. Image Process., vol. 29, pp. 710-723, 2020.
[7] A. Levin, Y. Weiss, F. Durand, and W. T. Freeman, "Efficient marginal likelihood optimization in blind deconvolution," in Proc. IEEE CVPR, Jun. 2011, pp. 2657-2664.
[8] L. Xu, S. Zheng, and J. Jia, "Unnatural L0 sparse representation for natural image deblurring," in Proc. IEEE CVPR, Jun. 2013, pp. 1107-1114.
[9] X. Chen, X. He, J. Yang, and Q. Wu, "An effective document image deblurring algorithm," in Proc. IEEE CVPR, Jun. 2011, pp. 369-376.
[10] X. Cao, W. Ren, W. Zuo, X. Guo, and H. Foroosh, "Scene text deblurring using text- specific multiscale dictionaries," IEEE Trans. Image Process., vol. 24, no. 4, pp. 1302-1314, Apr. 2015.
How to cite this paper
@article{1715080,
author = {P. Sujitha, S. Akshith Reddy, P. Nithisha, V. Srujan Naik, K. Chandana},
title = {Deblurring of Text Images},
journal = {Iconic Research And Engineering Journals},
year = {2026},
volume = {9},
number = {9},
pages = {970-975},
issn = {2456-8880},
url = {https://www.irejournals.com/formatedpaper/1715080.pdf},
abstract = {Blur in photographed or scanned text documents is a frequent and damaging degradation that reduces character legibility and OCR accuracy. This paper addresses blind text image deblurring, where neither the blur kernel nor its parameters are known in advance. The method builds a Text-Specific Hybrid Dictionary (THD) from three categories of patch pairs—Gaussian blur–sharp, motion blur–sharp, and sharp–sharp—that together cover the range of states an iterative solver encounters as it refines its latent image estimate. A Text Property (TP) Enhancement Operator, based on anchored regression on the trained dictionary, projects each intermediate patch toward the space of sharp text at every iteration. The overall objective is minimized by alternating between latent image and blur kernel subproblems using half-quadratic splitting. Final restoration uses a non-blind deconvolution step once the kernel is reliably estimated. Experiments on both synthetic and real-world blurred text images show consistent improvements over existing methods in PSNR, SSIM, and kernel similarity, with a corresponding gain in OCR accuracy. The experimental results demonstrate that the proposed approach effectively restores blurred text images and significantly improves text readability and OCR performance compared with existing techniques},
keywords = {Blind Text Image Deblurring, Hybrid Dictionary, Sparse Representation, Alternating Optimization, PSNR, SSIM, OCR Accuracy.},
month = {March},
doi = {https://doi.org/10.64388/IREV9I9-1715080}
}