High Capacity Steganography on Float-Point Number with Single Precision

High Capacity Steganography on Float-Point Number with Single Precision

연구 분야: Strategies

학회: VSIP '20: Proceedings of the 2020 2nd International Conference on Video, Signal and Image Processing

Nowadays, the research on modern image steganography mainly focuses on the spatial domain and jpeg transform domain. And the image elements that used to carry the message bits are all integers without exception. However, with the in-depth study of High Dynamic Range (HDR) images, single-precision float-point numbers have gained more and more attention. The float-point number has more redundant bits than integer, which makes it can hide more secret information and more suitable for steganography. In this paper, based on the IEEE754 standard, we employ the luminance channel of HDR image as the steganography domain which is named the mantissa spatial domain. Based on this new steganography domain, we propose a novel adaptive steganography scheme that can realize satisfactory security under high payload such as 0.75bpp. Experimental results show that the proposed scheme achieves better performance on resisting the state-of-the-art steganalysis method than existing schemes and make a good balance between capacity and security.