Restoration of Ancient Document Images Using Phase Based Binarization

The main defects present in historical documents are darkness, non-uniform clarification, bleed-through and faded characters. To remove these defects binarization method is used. In this paper a phase based binarization method is studied in which phase of ancient document images is preserved. This method is derived in to three steps: preprocessing, main binarization and post processing. In preprocessing phase preserved denoised image is derived. In main binarization two phase feature maps are derived are maximum moment of phase congruency covariance and a locally weighted mean phase angle. At last in post processing Gaussian and median filter is use for enhancement of image. It is also improve the performance of binarization methodologies.