An optical image watermarking algorithm, based on singular value decomposition (SVD) ghost imaging and multiple transforms, is designed. The watermark image is first encrypted by applying an SVD ghost imaging system, then the encrypted watermark is embedded into the cover image with the help of multiple transforms, including lifting wavelet transform (LWT), discrete cosine transform (DCT), discrete fractional angular transform (DFAT) and SVD. Four sub-band images are produced from the host image by LWT and DCT. The improved DFAT, whose scaling factors and parameter are optimized by particle swarm optimization algorithm, is operated in the new matrix. Afterwards, SVD is executed in the two-part image and the encrypted watermark is embedded in the host image by mutual operation of different matrices. Simulation results validate that the proposed watermark scheme is superior in the aspects of security, robustness and imperceptibility.

中文翻译：

---

基于奇异值分解鬼成像和粒子群优化算法的光学水印方案

设计了一种基于奇异值分解(SVD)重影成像和多重变换的光学图像水印算法。水印图像首先通过应用SVD重影成像系统进行加密，然后借助提升小波变换（LWT）、离散余弦变换（DCT）、离散分数角变换等多种变换将加密后的水印嵌入到覆盖图像中(DFAT) 和 SVD。通过 LWT 和 DCT 从宿主图像生成四个子带图像。改进的DFAT，其缩放因子和参数通过粒子群优化算法优化，在新矩阵中运行。之后，在两部分图像中执行SVD，通过不同矩阵的相互操作将加密水印嵌入到宿主图像中。