多光谱图像变换域统计量移位鲁棒无损信息隐藏

为了提高鲁棒无损信息隐藏算法容量，利用多光谱图像谱段间的相关性及其小波变换高频系数分块均值的稳定性，提出一种变换域统计量移位的鲁棒无损信息隐藏算法。首先，在光谱方向采用Karhunen Loêve变换(KLT)去相关，然后对得到的KLT本征子图像进行离散小波变换，将小波HH高频系数分块均值作为统计量，通过其直方图移位来隐藏秘密信息，最后将浮点数变换和像素溢出引起的误差嵌入含密图像中以便对恢复的载体图像进行误差补偿。实验结果表明：在图像质量和以往算法相当的情况下，该算法的隐藏容量增加了1倍，并且能抵抗噪声攻击；而与最新算法相比，在隐藏容量相同时，该算法抗JPEG2000压缩倍数平均提高23.9%，误码率平均降低了44.8%。

Multispectral image robust lossless data hiding based on transform domain statistical quantity shifting

In order to improve the hiding capacity,a transform domain statistical quantity shifting-based robust lossless data hiding scheme is proposed,which exploits the correlation between different bands of multispectral images and the stability of the block mean values of their wavelet transform coefficients. Firstly,Karhunen Loêve transform (KLT) is carried out to reduce the correlation between different bands,then discrete wavelet transform is conducted in the KLT obtained images,after that the HH coefficients are divided into blocks to calculate the block mean values,which are used as the statistical quantity,then their values are shifted to embed the secret data. Finally,the error values are embedded in the stego-image to compensate the loss caused by float transform and overflow. The comparison results indicate that the proposed scheme can embed twice the amounts of secret bits of that embedded in the previous methods when having the same image quality,additionally the proposed method can resist noise attack.Compared with the newest scheme by embedding the same bits of secret data,the proposed method achieves not only 23.9% increase of JPEG2000 resist compression ratio but also 44.8% reduction of bit error rate.