In this paper, an algorithm that embed watermarking into color QR code based on image normalization and contourlet transform is proposed, which is based on image normalization and invariant centroid theory. For the purpose of enhancing the invisibility and resistance to geometric attack of watermarking, we first encrypt the watermarking with chaotic method before information embedding, and then eliminate the effect of geometric change by utilizing image normalization. Based on the matrix singular value decomposition in contourlet domain, suggests an adaptive watermarking scheme that watermarks are embedded into the above resulted images. Corresponding, we perform inverse transform to the QR code which is attacked by geometric transformation and non geometric transformation including blurring, JPEG compression, noise addition, sharpening, scaling, rotation, and cropping before extracting the watermarking. Thus, we can prove the truth of the attacked QR code by the extracted watermarking which is recognizable.