Article Preview
Top1. Introduction
Nowadays, electronic documents such as contracts, military maps and engineering design drawings are generally used in many occasions. Among them, many of them are binary images. Meanwhile, to facilitate information exchange and sharing, many electronic documents are printed or copied in paper documents, and sometimes paper documents are also scanned into electronic documents for transmission in Internet. These provide chances for illegal users to steal or falsify these documents. Therefore, the copyright protection of the document after print-scan is a prominent problem to be solved.
Currently, digital watermarking was proved to be an effective means for copyright protection (Qi & Li, 2008). However, print-scan attack is still one of the most challenging issues for digital watermarking (Yu, Niu & Sun, 2005). The main reasons are:
- 1.
When an electronic document is printed and scanned, the regenerated electronic version may be the same as the original one from the naked eye, but some significant distortion may be produced, which is sufficient to destroy the embedded watermark.
- 2.
In print-scan operation, many parameters need to be manually set by users. In addition, the orientation of paper in printer’s paper input tray and on flat bed of scanner is also varied.
- 3.
After print-scan operation, the introduced distortion cannot be clearly distinguished, and no mathematical model and high precision device can accurately measure this distortion.
The above challenges impose high requirement on robustness of the corresponding watermarking scheme; otherwise, the embedded watermark cannot be successfully extracted. Meanwhile, binary image has little redundant space compared with gray or color image, so the distortion in it is more likely to be perceived by human vision system. Thus, robustness and image fidelity are two important targets of print-scan resilient watermarking for binary image.
Aim for the protection of paper map, a novel print-scan resilient watermarking based on flipping pixels, image scrambling and DCT of characteristic value matrix is proposed. The main contributions include:
- 1.
Image scrambling is applied to the watermarking scheme. It significantly increases the embedding capacity and makes it suitable for binary images with simple texture and uneven black area distribution, such as military maps and engineering graphics.
- 2.
The DCT of characteristic value matrix effectively solves the problem that the embedding watermark is removed after binarization operation, which enhances the robustness.
The rest of article is organized as follows: the related work is introduced in Section 2; the proposed print-scan resilient watermarking is described in Section 3; Experimental results and its comparative analysis are provided in Section 4; finally, some conclusions are drawn in Section 5.
TopComparing with grayscale or color image, the redundancy of binary image is small. Therefore, the researches of print-scan resilient watermarking for binary image is more difficult. Moreover, most of them are for text binary image (Qi & Li, 2008; Wu & Liu, 2004; Hou & Dai, 2015; Wang & Luo, 2016, He & Wu, 2009). According to the position for watermark embedding, the existed works can be classified into spatial domain based transform domain based print-scan resilient watermarking for binary image.