A novel hierarchical digital watermarking method, based on the properties of human visual system, is proposed. First the region watermark weights and local watermark weights are appropriately set based on the noise sensitivity properties of human visual system to different regions and local features. Watermark is then embedded into decomposed detail subbands of static wavelet transform. A new multiresolution detector is proposed for watermark detection. Experimental results and comparisons with other techniques operated in the wavelet domain prove the effectiveness of the new algorithm,including superior imperceptibility and robustness to common attacks such as image compression and median filtering.