An adaptive reversible watermarking in IWT domain

Abstract

An adaptive reversible watermarking in integer wavelet transform (IWT) domain was proposed in this paper. It could adaptively embed according to the size of the watermark, achieve blind extraction, and the extracted image was completely lossless. A new method of embedding watermark into diagonal components based on the relationship between adjacent elements in the approximate component was used. First, an integer wavelet transform was performed on the carrier image, and a threshold was adaptively set using the element relationship in the approximate component. Then a block with a higher smoothness in the approximate component was selected, and the watermark was embedded into its corresponding diagonal component image block. Since the threshold, the length and width of the watermark were also embedded in the image, it became possible to extract the watermark without knowing the threshold and the size of the watermark. Experimental data show that compared with different carrier pictures and similar algorithms, this method has better visual effects and less distortion, and the extracted image does not have any pixel difference from the original image. It can be used for special applications, such as medical image transmission, forensic document verification, etc.