Multiresolution based image adaptive watermarking scheme

This paper proposes an image adaptive watermark embedding method based on discrete wavelet transform (DWT). To increase the robustness and perceptual invisibility of watermark, the algorithm is combined with the quantization model based on human visual system (HVS). Number of factors that affect the noise sensitivity of human eye, such as background luminance, proximity to an edge, frequency band and texture masking are taken into consideration. To extract the embedded watermark, both blind (uncorrupted image is not required) and non-blind methods are introduced. Simulation results show that the proposed method is robust to the image intensity attack such as high compression.