Robust Spread Spectrum Binary Image Watermarking Based On Flipping Pixel

Watermarking mainly researches on how to embed special image, serial number or characters into digital media such as images, audio, video, without causing obvious visual loss. Watermarking is used for protecting copyright or distinguishing true or false. At present, watermarking primarily aims at gray or color images which have a large number of redundant information. Embedding watermarking into binary image is very hard because binary image lacks of redundant information. Now binary image has been widely used because of its simple storage and compact structure. A lot of important files, such as handwritten signature, fingerprint or document image, are stored and transferred in the form of binary images. Therefore, researching on binary image watermarking is of great significance for information security and application value.The main work of this paper is summarized as follows:Firstly, this paper introduces the background and significance of watermarking, such as system model, category, technical requirements and application. The technology difficulty of binary image watermarking is discussed. Typical algorithms of binary image watermarking are listed and contrasted. Evaluation indicators are introduced.Secondly, this paper implements a kind of watermarking algorithm of binary image based on flipping pixel. Spread spectrum is introduced to this algorithm because of its low robustness. In order to enhance efficiency of image patches, image scrambling is used to balance white pixel and black pixel. Meanwhile, a new pixel flipping model is proposed to improve visual quality. Experimental results show that improvement can be over pepper noise, Gaussian noise and cropping attacking. The BER of watermarking is lower than 10%.Thirdly, this paper proposes a new robustness watermarking algorithm. This algorithm embeds watermarking into image spatial domain based on flipping pixel. Watermarking can be extracted by symbol decision function. Experimental results show that this algorithm can be over against pepper noise, Gaussian noise and cropping attacking. The BER of watermarking is lower than 10%. The embedding rate of this algorithm is about 100 times that of the first algorithm on the similar image quality.