Watermarking Method Using Dual-Tree Complex Discrete Wavelet Transform and Quantization
Watermarking Method Using Dual-Tree Complex Discrete Wavelet Transform and Quantization
复制标题
DOI：
10.1109/icwapr56446.2022.9947102
复制
发表时间：
2022-09
期刊：
影响因子：
--
通讯作者：
Masayuki Shimoshimbara;Ken Muraoka;Teruya Minamoto
中科院分区：
文献类型：
--
作者：
Masayuki Shimoshimbara;Ken Muraoka;Teruya Minamoto
近年来，随着信息社会的发展，数字图像可以很容易地上传和下载。在这种背景下，剽窃和篡改图像的情况一直在增加。因此，有必要研究鲁棒性水印技术来实现数字图像的版权保护。本文提出了一种基于双树复离散小波变换和量化的数字图像盲水印方法。我们将此小波变换应用于原始图像，并使用量化索引调制和相对调制嵌入两个不同大小的水印图像。量化索引调制对几何攻击具有鲁棒性。相对调制对噪声添加是鲁棒的。实验结果表明，该方法对高斯噪声、椒盐噪声、剪切和JPEG压缩具有较好的鲁棒性。实验结果表明，该算法能较好地抵抗高斯噪声、椒盐噪声、剪切、JPEG压缩等噪声干扰。此外，我们视觉上检查提取的水印图像。此外，我们还比较了三种方法：一种是采用双树复离散小波变换和滑动窗口。第二种方法是基于离散小波变换，离散余弦变换，奇异值分解。第三种方法是基于离散余弦变换和RM，QIM。通过与现有方法的比较，验证了本文方法的有效性和有效性。
In recent years, with the development of the information society, digital images can be easily uploaded and downloaded. Against this background, instances of plagiarism and tampering with images have been increasing. Therefore, it is necessary to develop robust watermarking for the copyright protection of digital images. In this study, we propose a blind digital image watermarking method based on dual-tree complex discrete wavelet transform and quantization. We apply this wavelet transform to the original image and embed two watermarked images of different sizes using quantization index modulation and relative modulation. Quantization index modulation is robust against geometric attacks. Relative modulation is robust against noise addition. The results of our experiments show that our method is robust against Gaussian noise, salt and pepper noise, cropping, and Jpeg compression. The experimental results show that the watermark can be extracted by Gaussian noise, salt and pepper noise, cropping, Jpeg compression. In addition, we visually check the extracted watermarked images. Furthermore, we compare three methods: one using dual-tree complex discrete wavelet transform and sliding window. The second method is based on discrete wavelet transform, discrete cosine transform, and singular value decomposition. The third method is based on Discrete Cosine Transform and RM, QIM. By comparing these three methods with the existing methods, we confirmed the significance and performance of our method.