A Proposed Optimum Threshold Level for Document Image Binarization

Abstract

Today, document binarization is very important especially for text analysis. In background degradation, it’s complicated to determine the optimum threshold. The inappropriate threshold value will be effected on the segmentation result, such as produced artefact and black noise. In this study, a modified algorithm of the Nick method in order to obtain the optimum threshold value was presented. The aim of this work is to improve the performance of Nick method and achieved a better binarization result compared to the other existing binarization methods. The results of the numerical simulation indicate that the proposed method obtained highest in terms of F-measure (71.399) and the PSNR (14.662) compared to the Otsu method, Niblack method, Bernsen method, and Nick method. Conclusion, the proposed method capable and effective to deal with the low contrast and black noise problem.