No abstract

No abstract

In this paper, we propose a novel hybrid mixed frequency domain scheme. The secret message bits are embedding on the singular values of the blocks within low frequency sub-band in host digital image. Using low-frequency approximation sub-graph transformation coefficients restore high-frequency element sub-images transformation coefficients. When detecting secure message, by comparing replaced part coefficient we can find out whether the image has been tamper. Simulation results are evidence for that the algorithm has strong detection and location capability and maintain better original image quality. To embed secure message imperceptibly, robustly and securely, we model the adaptive quantization parameters by considering the human visual system (HVS) distinctiveness. Experimental results demonstrate that the proposed scheme is robust to variety of attacks. Secure message extraction is efficient and blind in the sense only quantization strategies but not the host image is -required.