Restoration of Degraded Historical Document Image: An Adaptive Multilayer-Information Binarization Technique

Binary image is the essential format for document image processing, and the operation of the subsequent steps depends on the quality of the binarization process. The objective of this research is to propose a new binarization method based on adaptive multilayer-information for restoration of degraded historical document images. This paper focuses on degraded Thai historical document images, which are in the form of handwritten and machine-printed documents images. The proposed method consists of five stages including noise elimination, majority pixel analysis, degradation of the background layer estimation, thresholding and vicinity analysis. The experiments are performed on 480 degraded Thai historical document images provided by National Library of Thailand. The experimental results demonstrate that the proposed method performs better than five well-known adaptive binarization methods.