Research on Digital Image Watermarking
|School||Harbin Engineering University|
|Course||Communication and Information System|
|Keywords||watermarking robustness Human Visual System(HVS) geometrical attacks channel capacity|
With the development of digital multimedia technology, digital watermark has become a popular topic in the field of digital copyright protection. Digital watermark servers as protection the securities of digital multimedia by embedding special information into it. Applications of watermarking include not only copyright protection, but also document authentication, covert communications and data embedding, etc. In these applications, a watermark is embedded within a host data. This embedding should be nearly imperceptible and robust against possible manipulations of the watermarked data, in the sense that it should be possible to reliably extract the watermark in a degraded version of the signal. Degradations include operations such as addition of noise, filtering, compression and geometrics distortion. These degradations could be intentional (due to an adversary) or nonintentional (e.g. due to a common image processing). The designing of a watermark resistant to these distortions is still an open problem. Robustness is crucial for watermarking technology in order to be used widely in commercial areas.This paper takes still images as example, systematically studied the technology of digital watermarking. The frameworks and methods of embedding, detection and watermark signal encoding were presented. Theoretical analyses of the watermark detector’s statistical characteristics and the channel capacity was presented. These are helpful for development efficient watermarking schemes.This paper focused on the study of watermarking schemes in frequency domain and tried to develop robust watermarking methods. A DCT domain blind watermark scheme was proposed, which is image self-adaptive and taking into account the HVS. Three masking factors of luminance, frequency and texture were formulated according to the local characteristics of the host image. These three factors were calculated directly according to the 8x8 DCT blocks coefficient.