Enhancement of Palm-Leaf Manuscript and Color Document Images with Synthetic Background Generation

This paper presents a new color document image enhancement technique that is suitable for both palm-leaf manuscripts and color document images. The proposed technique consists of three steps: preprocessing using Wiener filter, contrast enhancement and background enhancement, binarization using new global thresholding technique and post processing for synthetic background generation.

In the preprocessing stage, we propose three sub-stages. The first sub stage is used to eliminate the noisy areas and smooth out the background. Second sub-stage enhances the contrast between background and text areas where as the third sub-stage takes care of non-uniform background. Binarization is performed by using new global thresholding technique. This technique helps in properly extracting useful text information from the low quality document and palm-leaf manuscript images. Finally, the post-processing stage resynthesizes the color background. The proposed technique has been tested on a set of palm-leaf manuscript and color document images having various deformities such as uneven background color, folds, character breaks and localized spots.