Abstract: This paper proposed a new digital watermarking algorithm based on compressive sensing and wavelet transform,for copyright protection.At first,sparse decomposition with wavelet transformation is carried out on the original image,and the low-frequency coefficients is scramble with Arnold scrambling.Then Gaussian random observation matrix is applied to implement compressive sensing measurement on the coefficient matrix.The watermark is embedded into the perceptual domain,namely,the compressed signal.At last,The OMP algorithm was adopted to restore the recomposed low-frequency matrix signal.And the image is reconstructed with wavelet transform.In this watermarking algorithm,the measurement matrix is also served as the key,thus enforce the security.In addition,Arnold scrambling increase the invisibility and security of the watermark.The simulation results show that this new algorithm is robust to noise,filtering and rotation attacks.