Robust JPEG Image Steganography Using Wavelet Domain SVD and Adaptive QIM

Massive images in Internet networks provides rich cover objects for covert communication using digital steganography. However, the lossy processing such as image compressing and image scaling often take place in some network platforms, which makes the existing image steganography algorithms designed for lossless environment unusable for lossy network platforms. Based on singular value decomposition (SVD) in stationary wavelet domain, a robust and adaptive image steganography algorithm is proposed, which can resist lossy image compression and extract the embedded secret message correctly. First, stationary wavelet transformation is performed for cover image, and the low-frequency subband is divided into non-overlapping blocks, and SVD is performed for each block; second, considering the stability of maximum singular value, the cover elements are extracted by applying a adaptive quantization index modulation (QIM) process on the maximum singular values; third, the embedding distortion function is defined according to the influence of the cover element change on the inter-block and intra-block wavelet coefficients, and the secret messages are embedded using syndrome-trellis code (STCs); finally, the adaptive QIM is used to embed the secret elements and generate the stego image. The experimental results show that the proposed steganography algorithm has better robustness and stronger anti-detection capabilities.