Abstract: In order to solve the contradiction between image hiding imperceptibility and robustness in the NSCT, a method of hiding visible light image is proposed based on combination of wavelet packet transform and NSCT, which uses NSCT transformation to decompose the carrier image into a set of low-frequency sub-band and high frequency sub-band, and then to decompose low-frequency sub-band with two level wavelet packet. The method can ensure the secret key bit-plane image information hidden in the wavelet low-frequency sub-band by singular values transform, and the secondary information adaptively hidden in high frequency sub-band of NSCT. The experiment shows that, in the same embedding capacity, peak signal to noise ratio of this algorithm is greater than 50 dB, after geometric attacks and filtering processing of the interference, the secret image normalization coefficient is still close to 1.

Key words: digital image, image hiding, Nonsubsampled Contourlet Transform（NSCT）, Wavelet Packets Transform（WPT）, Sigular Value Decomposition（SVD）

摘要： 为了更好地解决NSCT域图像隐藏不可见性和鲁棒性之间的矛盾，提出了一种基于NSCT变换和小波包变换相结合的可见光图像隐藏方法，利用NSCT变换将载体图像分解为低频子带和一组高频子带，对低频子带进行二级小波包分解，通过奇异值变换将秘密图像重要位平面信息隐藏在小波包分解低频子带中，次要信息自适应隐藏在NSCT高频子带中。实验表明，在同等嵌入容量下，算法峰值信噪比大于50 dB，对几何攻击和滤波等干扰处理后，秘密图像的归一化系数仍接近于1。

关键词: 数字图像, 图像隐藏, 非下采样Contourlet变换（NSCT）, 小波包变换, 奇异值分解