Details of Research Outputs

In Voice over IP scenarios, adaptive partial-matching steganography (APMS) is an effective approach to balance steganographic transparency and bandwidth, in which the strategies involved, however, have not achieved the optimal steganographic performance yet. Therefore, in this paper, we try to improve APMS from three aspects. First, when the current partial similarity value (PSV) falls between the two given substitution thresholds, the embedding process is guided with unequal probabilities increased with the PSVs instead of identical probabilities, which enhances the embedding chances of cover parts with larger PSVs. Second, for cover parts considered unsuitable for information hiding in the original strategies, the matrix encoding strategy is employed to conceal secret messages with no more than one bit changed, which improves cover usage rate while maintaining high embedding efficiency. Third, in the APMS strategy with two flag bits, one more encrypted form of the secret message is added into the comprehensive evaluation of partial matching, which makes the embedding process more efficient. The proposed strategies are evaluated with ITU-T G.711 (A-law) as the codec of the cover speech, and compared with the original ones and some existing approaches. The experimental results demonstrate that the improved strategies can strike excellent balances between the steganographic transparency and bandwidth, and significantly outperform both the former ones and other previous approaches. Furthermore, the results of detection tests show they have outstanding performance of steganalysis resistance.