In order to enhance the resistance of QR code watermark image against geometric attack, the paper proposed a new watermarking algorithm based on singular value decomposition and discrete wavelet transform. The text information was transformed to host image by the principle of encoding; then the logarithmic coordinates transform was applied to the host image. Afterwards, the scrambled watermark was embedded into the low-frequency sub-block of the host image through discrete wavelet transform using singular value decomposition principle, forming the watermarked QR code. Experiments showed that this algorithm could make the QR codes watermark image resist rotation, scaling and translation (RST), and other geometric distortions. The algorithm had good robustness and invisibility, and can thus serve as a viable copyright protection and authentication tool.