At present, those existing watermarking algorithms for print forgery prevention always embed watermarks into original images, and their invisibility needed to be improved. Moreover, they had never studied their robustness against dual print-and-scan process. A watermarking algorithm for print forgery prevention was proposed to overcome those problems based on the comparison of numerical relationship between norm and mean of norm. It conducted discreet wavelet transformation on the original image, split its low frequency band into non-overlapping blocks, and conducted singular value decomposition on each block. The zero-watermark sequence was derived from judging the numerical relationship between singular matrix’s norm from each block and mean of singular matrix’s norm form all blocks. It has perfect visibility due to no alteration made to the original image. The experimental results show that it has the ability to resist both single print-and-scan process and dual print-and-scan process.