Performance comparison of progressive QIM and QIM using different wavelet families

In this work, we investigate the performance of two watermarking scheme QIM and progressive QIM through a statistical analysis. We rely on imperceptibility and robustness as a measure of performance of a watermarking algorithm. Seven wavelet families are used to generate the carrier signal. We insert the watermark message in approximation as well as in different subbands, at a different level of transformation in each wavelet, one at a time. To test the robustness, the watermarked image was subject to several attacks: JPEG, gamma correction, median filter and Gaussian noise.