基于DCT的多门限渐进秘密图像分存方案

Multiple Thresholds Progressive Secret Image Sharing Scheme Based on DCT

图像分存是将密图拆分为影子图像,利用分发的部分或全部影子图像对密图进行完整重构,是目前图像信息安全的研究热点。但传统数字图像分存通常是由满足门限数量的影子图像对密图进行完整重构的,而在频域随着分发份额数量增加,实现对密图由模糊到清晰的整体渐进恢复的研究工作相对较少。基于此,文章提出一种基于DCT的多门限渐进秘密图像分存方案,在该方案中,首先结合随机参与值将密钥分存,将分发子密钥和参与值对应的MD5值公布防止参与者作弊;然后对秘密图像进行8×8DCT变换,对分块DCT系数进行随机量化和指定码长分配表的2进制表示,按之字形扫描顺序和频带重组形成多个划分频带;最后在GF(2~8)有限域下对频带进行多门限分存并重构分发影子图像。实验结果表明,当参与者的数量大于最低门限,则随着参与者数量的增加,文中方法可充分利用影子图像对秘密图像进行由模糊到清晰的渐进恢复且具备较好的视觉恢复质量,且得不到秘密图像的任何信息。文中策略严格依赖于密钥,并且恢复的结果趋于稳定。

Image sharing usually splits the secret image into a series ofshadow images and the secret image can be completely recovered by partial or all distributed shadow images. It is the research hotspot in image security. But traditional image sharing usually used to construct the secret imagecompletely and the research providingthe whole progressive recovery of secret image from fuzzy to clear is relatively small.Based on this, a multiple thresholds progressive secret image sharing scheme based on DCT was proposed.First the key is shared in combination with the random participation value,the MD5 values corresponding to the sub-keys and participating values are published to prevent participants from cheating,then the secret image is transformed by 8 x8 DCT where the DCT coefficients are randomly quantified and the binary representation of the assigned code length distribution table is presented to form multiple partition bands by zigzag scanning order and band reorganization.Finally, use multiple thresholds to sharing the band and reconstruct distribution shadow image in GF（2~8） Galois field.The experimental results show that when the number of participants is greater than the minimum threshold, withthe number ofparticipant＇sincreases, the proposed method can make use of the shadow image to make the whole progressive recovery of secret image from fuzzy to clearand the more participantsand the better visual recovery quality. Otherwise, there will not be any information on the secret image, and the strategy is strictly dependent on the key and the recovery results tend to be stable.