ROI Digital Watermarking Based On The Texture Characteristics

With the rapid development of computer network and multimedia technique, digital media, such as image, video, sound and so on, can get copied, spread, processed and brought into the public easily with the help of internet. This makes people doubt about the integrity and authenticity of the digital media. Digital watermarking technique can overcome the traditional encryption and digital signature technique’s deficiency. So as a new information hiding technique, it has become an effective protection approach in the open network environment. Authentication watermarking has been the most effective method to ensure trustworthiness of multimedia data, and it has been developed and become a hot research field. Semi-fragile watermarking technique is an important branch of authentication watermarking. It is robust for common signal processing operations, while fragile for intent changing manipulations. It can distinguish from general image processing and tampered image. This paper does research on semi-fragile authentication watermarking which is used for protecting the integrity and authenticity of image.At present, most of the watermarking algorithms view the original image as a whole, not taking the important features of the original’s own and some other specific location into account. However, some given area is more significant than other ones for many applications. Take the image with a person in the middle for example, generally speaking, the appreciators or the author will pay more attention to the person in the image than the background. The so-called region of interest (ROI) is the attractive region. This paper studies up on image’s ROI, and the algorithm proposed can have key protection to ROI.In addition, this paper also makes a research on characteristics of image. It apply texture feature to digital watermarking algorithm and have in-depth discussion of it. Among the numerous textural feature extraction methods, the paper chooses the classic one named GLCM. As there are many parameters in this method, this paper provides the principle of choosing some of them as texture feature through theoretical analysis and experimental verification.On the basis of summarizing and analyzing to relative research at home and abroad, combined with textural feature and ROI theory, this paper proposes a digital watermarking algorithm of ROI based textural feature in order to overcome the deficiencies mentioned above. The algorithm adopts the most frequently used linear transformation named Discrete Cosine Transformation (DCT) with considering the feature and property of the image’s own, which can control the watermarking embedment strength according to the complexity of textural feature. It is proved that not only it can guarantee the transparency, but also enhance the robustness of watermarking. Moreover, this algorithm uses spread spectrum and watermark scrambling technique during the generative process to improve watermarking security observably. As a semi-fragile watermarking technique, the algorithm not only can put up with a certain degree of normal signal processing, but also can detect malicious modification to the multimedia data. Under the action of ROI masking, the proposed algorithm can make accurate positioning to the tampered area in ROI, but have no action to the region of uninterested. In the end, a mass of experiments verify the proposed algorithm’s feasibility and effectiveness.