Digital Watermarking Algorithms For Halftone Images

Since its emergence ten years ago, digital watermarking has been widely used inareas such as copyright protection, piracy tracing, copy control and imageauthentication, and more and more researchers have paid their interests in it.According to the media used to embed watermark, digital watermarking techniquescan be classified into different categories. In this thesis, we mainly focus onwatermarking in halftone images.Halftone watermarking can be further categorized as robust halftonewatermarking and fragile halftone watermarking, each of which has its own specificapplications. Robust watermarking, since being robust to common image processingor manipulation, appeals to copy protection, piracy tracing and copy control. In thecontrast to robust watermarking, fragile watermarking is very sensitive to imageprocessing, so it is a good option for image integrity authentication. Here, we dip intoboth of them.In the area of robust halftone watermarking, we first introduce an existedblue-noise watermarking scheme, which has good robustness, and can stand strongrandom image noises. The shortcoming of the method is that in dark images, itsrobustness is not as good as in usual images, and under non-ideal printing conditions,the output of the approach will suffer great distortion. We offer an improvedembedding method and a green-noise watermarking scheme, which effectivelyenhance the original method.As far as fragile halftone watermarkings, we introduce an authenticationwatermarking scheme which embeds watermark in random locations of an image. Thescheme can detect pixel-level modifications, but due to the random choice ofwatermarking locations, its visual effect is image-dependent, and to some images, theoutput is not appeal. We use HVS model to guide the decision of watermarking places,what's more, we present a novel halftone method to generate more candidatewatermarking locations. Experimental results show that the measure we take isdramatically effective.