Digital Watermarking Based on Quantizing Discrete Cosine Transform Coefficients of Still Image

摘要

To provide copyright protection for digital media, digital watermarking is proposed and has been drawn extensive attention. A new technique based on quantization discrete cosine transform coefficients of a still image to embed a digital watermark into a still image is proposed in this paper. This technique does not require the original image when extracting watermark from suspicious image. During watermarking, the original image is blocked according to the model of the Joint Photographic Expert Group (JPEG) lossy compression. And also the pseudo-random permutation is used to improve the robustness to cropping operation. To make it clear, firstly, this paper introduces the principle of coefficient quantization. Secondly, the method of embedding watermark into the still image based on quantizing discrete cosine transform (DCT) coefficients is discussed in detail. Thirdly, the method of extracting watermark from a suspicious image is discussed. At last, several experimental results demonstrate that the embedded watermark is invisible and robust to most of the signal processing operations, such as filtering, cropping, lossy compression, and so on.