Watmerking Alogorithms And Transparency Assessment For Digital Images

Digital watermarking embeds digitized information into multimedia carriers such as texts, audios, images and etc. Watermark embedding will not affect the commercial value and useness of original media and the watermark is detectable and extractable. Multimedia could act as objects being protected or carriers for secret communication. Let digital image watermarking as research content, this paper focuses on the following aspects and creativeness.1. Proposes a fundamental watermarking model based on general vector quantization. Vector quantization is formed according to Shannon's rate-distortion theory, which could be regarded as a combination of one encoder and one decoder. This model could interpret most watermarking algorithms even zero watermark technique. When the model is applied, the capacity of the algorithms could be estimated accurately. The model implies that the capacity of watermarking is up to codebook construction and grouping.2. For watermarking transparency consideration, proposes a visual quality assessment based on Euclidean distance special for binary images. It is difficult for traditional methods to assess binary images'quality objectively. Quality evaluation specially for binary images proposed here, utilizes Euler number to describe structure distortion, measures noise energy by distance and piecewise energy function between noise and signal points, weights noise energy by structural effect factor decided by how noise influences the signal structure, and the same time considers the mass effect of noise points. Finally it turns out a quantitative analysis for images being tested.3. For watermarking transparency consideration, a quality assessment for grayed images, SNPSNR, based on multiscale textural structure and normalized noise. By virtue of the Gray-level Difference Statistics and time-frequency property of...