Restoring severely blurred and degraded text document images remains a challenge, particularly under non-uniform spatial blur and illumination conditions. In this study, we propose a robust image enhancement pipeline to restore and binarize text documents affected by varying degrees of blur. The method integrates Richardson-Lucy deblurring, frequency with Gaussian point spread function and spatial domain blur estimation, morphological filtering, and an adaptive thresholding scheme. To evaluate its effectiveness, the proposed method is compared with Sauvola, Niblack, Wolf, Bernsen, proposed + Global thresholding, and Richardson-Lucy alone, across two levels of degradation scenarios. Quantitative analysis uses F-Measure, PSNR, SSIM, Misclassification Error (ME), and Negative Predictive Measure (NPM). Experiments are applied on a dataset of 417 text images to demonstrate the effectiveness of the proposed method under level one and level two conditions of blur. Combining the proposed method with global binarization showed reliable results, although with reduced text detail restoration. Experimental outcomes and statistical analyses further validated the robustness and stability of the proposed method. The method's simplicity and adaptability make it suitable for document pre-processing in archival, legal, and OCR-driven applications.
Karthik et al. (2026) studied this question.