In this study we proposed a novel blind digital watermarking algorithm in wavelet domain. In order to realize extracting which not to depend on the original image excessively, we apply quantifying idea. Since the human visual system has especial responses on textures, the watermark embedded on textures has good robustness to general image processing and other attacks. So adjusting embedding strength through classifying embedded coefficients by using the detail coefficients that lies in the same layer with embedded coefficients. Present algorithm proved its robustness to general image processing and geometric attacks through experiments. The conclusions obtained by experiments are useful to the copyright protection and covert communication research in the future.