Fast and Secure Steganography Based on J-UNIWARD
Ante Su, Sai Ma, Xianfeng Zhao
Abstract
Adaptive steganography based on minimizing additive distortion model with Syndrome-Trellis Codes (STCs) is considered as a state-of-the-art technique of covert communication. However, the running time on mobile phone with large size image is not acceptable. The current approach only accelerates STCs embedding. J-UNIWARD, as one of the most secure distortion function, its calculation time is ten times longer than STCs embedding time. Therefore, just accelerating STCs embedding will only result in a small decrease in total running time. In this letter, we propose a fast and secure steganography based on properly simplifying J-UNIWARD. A simplified distortion function based on J-UNIWARD is designed to reduce time complexity. Besides, we propose a segmented STCs, which increases the computational parallelism of STCs embedding by several times without significantly degrading the security. Experimental results demonstrate that the proposed method can be three times faster than J-UNIWARD at only a very slight price of security.