Abstract A secure, robust, and blind watermarking method based on computational ghost imaging (CGI) in the discrete wavelet transform domain is proposed. In this method, an original watermark is firstly encrypted by the CGI system with chaotic key, then the sequences of the ciphertext are quantified into two levels and hidden in the wavelet coefficients of a host image. The quantization on the ciphertext compresses the amount of data to be hidden but does not induce serious deterioration in the retrieved image comparing with the conventional CGI. Simulation results show that the decrypted watermark is still distinguishable and can be authenticated with the nonlinear correlation coefficient. The embedded data can be blindly extracted with no help of the original host image. The chaotic key acts as both the encryption and the embedding key, which guarantee the security of the watermarking scheme.

中文翻译：

---

一种基于小波域计算鬼成像的盲水印方案

摘要 提出了一种基于离散小波变换域计算鬼成像(CGI)的安全、鲁棒、盲水印方法。该方法首先通过CGI系统用混沌密钥对原始水印进行加密，然后将密文序列量化为两级并隐藏在宿主图像的小波系数中。密文的量化压缩了要隐藏的数据量，但与传统的 CGI 相比，不会导致检索到的图像严重恶化。仿真结果表明，解密后的水印仍然是可区分的，可以通过非线性相关系数进行鉴别。嵌入的数据可以在不借助原始宿主图像的情况下盲目提取。混沌密钥既是加密密钥又是嵌入密钥，