为了提高YASS 算法的隐写安全性,提出一种基于灵活次序的虚拟嵌入块和量化器原点随机抖动的YASS 改进算法——VHD-YASS 算法. 为了增加嵌入位置的随机性,该改进算法在每个载体块内随机选取一个不规则区域,重新生成一个虚拟的8×8 小块来取代传统YASS 算法中规则的8×8 嵌入块,用于消息的嵌入;然后,通过密钥控制量化索引调制( QIM) 中奇/ 偶量化器的原点偏移,消除QIM 量化步长较大时消息嵌入造成的DCT 系数聚集效应. 将文中改进算法与传统YASS 算法和两种不同的改进YASS 算法进行了对比实验,结果表明:在同一嵌入容量下,文中改进算法的检测概率最高为0. 614,而传统YASS 算法的检测概率最高为0. 983. 从抗隐写分析能力、可视性及算法复杂度方面综合分析可知,文中算法是一种有效的隐写算法.

In order to improve the security of YASS algorithm, a modified high-randomness YASS algorithm namedVHD-YASS is proposed based on the virtual embedding blocks and the origin dithering of quantizers. In this algo-rithm, in order to improve the randomness of the embedding position, a virtual 8×8 H-block, which is used asa substitution of regular 8×8 H-block in the conventional YASS algorithm for message embedding, is created byrandomly selecting an irregular region in each B-block. Then, the shifts of origins of odd/ even QIM (Quantiza-tion Index Modulation) quantizers are controlled by a secrete key to avoid the DCT coefficient clusters due to themessage embedding under the condition of long QIM step. Finally, the proposed algorithm is compared with theconventional YASS algorithm and with two other improved YASS algorithms by experiments. The results show that,at the same embedding rate, the proposed algorithm attains a maximum detectability of 0.614, which is dramaticallyless than the conventional one of 0. 983, and that VHD-YASS is an effective steganography scheme because it is oflarge anti-steganalysis capability, excellent visibility and low complexity.