Traditional digital watermarking in transform domain computes complex and has limited anti-attack. Different ways of signal sparsity of compressed sensing represent different domains, which expands the embeddable space of watermarking. To take advantage of its multi-direction and multi-resolution characteristics to sparse image, Contourlet transform was analyzed. For the limitation of traditional block decision, we adopt the interleaving extraction to reduce the block artifacts. Furthermore, we presented the watermarking algorithm based on interleaving extraction block compressed sensing in Contourlet domain, simulation results verifies that the algorithm enhances the peak signal to noise ratio. Attack tests show that the proposed digital watermarking has better robustness against JPEG compression, noising, cutting etc.