An Adaptive Hybrid Algorithm for Document Images Binarization subject to Complex Background

Abstract

In this thesis, we introduce two adaptive and hybrid binarization methods. The first method is proposed for solving all types of degra- dation, such as non-uniform background, faint text, low contrast, stain, bleed-through, and shadow. This method depends on Otsu multilevel thresholding method, the estimated stock width, and the image contrast. The second proposed method aims to binarize the document images contain normal illumination using an improved version of the spectral clustering algorithm in speed and complexity. The thesis is divided into six chapters as listed below: