The work aims to further improve the embedding capacity and robustness of blind watermarking algorithm, and propose a robust double color blind watermarking algorithm based on QR code and DWT-SVD. Firstly, the color QR code as watermark was scrambled and normalized in each channel based on Arnold transmission, then each channel of carrier image was processed by the wavelet transform and the level 3 low frequency sub-band was partitioned into 4*4 blocks for SVD decomposition; finally, the watermark information was embedded into the corresponding channel of carrier image by adjusting the difference of (2, 1) and (3, 1) coordinate coefficients of each sub-band’s unitary matrix. The experiment results showed that the algorithm had high robustness, and it had strong ability in resisting such attracts as cutting, rotation, JPEG compression, Gaussian noise, salt and pepper noise, Gaussian filtering, median filtering and contrast enhancement. Because the algorithm does not need any information related to the original carrier image during the extraction process, it is a blind watermarking algorithm. Therefore, the proposed double color blind watermarking algorithm can be widely used in copyright protection of digital products.