The paper aims to improve the robustness and invisibility of the digital watermarking, image blind watermarking algorithm based on Shamir threshold scheme and DWT is proposed. Firstly, after Arnold transformation of the watermark image was stored with Shamir threshold scheme. Then, the original image was divided into blocks and decomposed into two levels of wavelet, the corresponding low frequency coefficients were extracted, and then the separated watermark was embedded into the low frequency coefficients of the corresponding original image blocks by overlay method. Finally, the synthesized image completed watermarking embedding. The experimental results show that the algorithm has good invisibility, PSNR is above 49 dB, SSIM is close to 1, and the NC value of extracted watermarking is higher than 0.9 under various attacks. The algorithm demonstrates good robustness to shear, JPEG compression and common noise jamming attacks. And the extraction of the watermark does not need the original image. The algorithm achieves the blind extraction of the watermark and has realistic feasibility in copyright protection.