Highly reliable digital watermarking using successive subband quantization and human visual system

A wavelet-based digital watermarking algorithm is proposed that uses successive subband quantization (SSQ) and the human visual system (HVS). First, the original image is decomposed into four levels using a discrete wavelet transform (DWT), then a watermark is embedded into the perceptually significant coefficients (PSCs) of the image. The coefficients on the lowest level are excluded in the watermark embedding step as these coefficients can be easily eliminated and modified by lossy compression. The PSCs in the baseband and other high frequency subbands are selected according to the amplitude of the coefficients and by SSQ, respectively The embedding of the watermark follows two different paths based on whether the selected PSCs are located in the baseband or high frequency subbands. As such, a constantly weighted watermark, which has a largest value without the introduction of a visual artifact, is embedded into the PSCs in the baseband. Whereas, for the PSCs in the high frequency subbands, the watermark is embedded based on the HVS so as to provide invisibility and robustness. In computer experiments, the proposed watermark algorithm is found to be superior in invisibility and robustness to the conventional algorithm of Dugad et al. (1998)