Abstract: A phase-based binarization model is applied for the binarization of an ancient document images , in addition a post-processing method is used to improve any binarization method. From the phase information of an input document image three features are derived, these features constitute the core of this binarization model. The features are the maximum moment of phase congruency covariance, a locally weighted mean phase angle, and a phase preserved denoised image. Three standard steps that are used in the paper are 1) pre-processing, 2) main binarization, and 3) post-processing. In first two steps, the features used are mainly phase-derived, while in the post-processing step, specialized adaptive Gaussian and median filters are considered. Outputs of the binarization step, which have high recall performance, is used as input to a post-processing method to improve other binarization methodologies. The experimental results on the different data sets, that is DIBCO, PHIBD, and BICKLEY DIARY show the robustness of the proposed binarization method on different types of degradation.
Keywords: binarization, phase-derived features, ground truth, document enhancement.