Restoration of stained old manuscripts via a hybrid wavelet and bilateral filtering system

Abstract

The conservation and restoration of old stained manuscripts is an activity devoted to the preservation and protection of things of historical and personal significance made mainly from paper, parchment, and skin. We present in this paper a hybrid implementation for de-noising and restoration of old degraded and stained manuscripts. This implementation is based on the statistical dependence of the wavelet coefficients of type Ortho-normal Wavelet Thresholding Algorithm based on the principle of Stein’s Unbiased Risk-Estimate Linear Expansion of Thresholds (OWT SURE-LET) and the synergy with bilateral filtering. First, the non-biased quadratic risk Stein estimator is applied to de-noise images corrupted by white Gaussian noise. In a second step, an improved bilateral filter is introduced to smooth and eliminate unnecessary details with the advantage of preserving edges between image regions. Obtained results show the effectiveness of the proposed synergy compared to separated approaches both on gray scale images and stained old manuscript.