关键词: 可逆信息隐藏, 高保真, 奇异值分解（SVD）

Abstract: In order to solve the problem of distortion caused by embedding the secret data on the original carrier image, high-fidelity reversible information hiding algorithm using Singular Value Decomposition（SVD） prediction pixel is proposed in this paper. Firstly, the original carrier is divided into gray and white layers. Pixels in the gray layer are selected as the target pixels, and the white pixels in the field are used as the reference pixels. Then, the neighborhood matrix is constructed using these reference pixels, the matrix is processed using SVD, and is restored with a large singular value, the target pixel is predicted by its average. Finally, the secret data is embedded by the extended prediction error. Extensive experiments have shown that the proposed algorithm effectively reduces the embedded distortion of the carrier.

Key words: reversible data hiding, high fidelity, Singular Value Decomposition（SVD）