The work aims to improve the ability of anti-signal processing attacks of digital watermarking algorithm, in order to improve the watermark robustness. A watermarking algorithm based on Contourlet transform and singular value decomposition (SVD) was proposed. The watermark image was repeatedly subject to Arnold scrambling processing to achieve scrambling encryption; directional sub-bands of Contourlet transform coefficients were selected as the watermark embedding region; SVD decomposition was conducted in the Contourlet coefficient building matrix and the adaptive selection of embedding strength was done combined with the human visual system. The experimental results showed that the peak signal to noise ratio of the image was 54.1279 dB, which indicated that the proposed algorithm had good transparency and concealment. The proposed image digital watermarking algorithm has good robustness, invisibility and the ability to resist various attacks, and it promotes the copyright protection of digital products.