Local Adaptive Thresholding Techniques for Binarizing Scanned Lampung Aksara Document Images

Abstract

Lampung characters are one of the region-heritage characters from Indonesia. However, the current trend makes these characters becoming unknown. The concern about the extinction of these characters makes several researchers tried to digitized the documents which contained Lampung Character. Nevertheless, the process of digitalization is not free from noise. This problem makes us want to handle the noise from the scanned documents using binarization and noise removal techniques, especially in Lampung characters' documents. In this paper, we implemented local adaptive thresholding using the Niblack method and the Sauvola method for thresholding value. We also implemented Adaptive Wavelet Thresholding for Bayes Shrink for removing salt and pepper noise from the binarization process. The result showed that the Sauvola thresholding gives better results compared to Niblack thresholding. Our contribution in this paper is the implementation of both processes in Lampung Characters document