The digital watermarking technology is a new information encryption technology. It has a good capacity in camouflage. However, it is limited by the phase detection errors, and the traditional retrieval algorithms are unpractical by means of calculating phase first and then deducing the original images in the process of restoring signal. The iteration algorithm can efficiently reconstruct these weak signals by applying some constrains in frequency domain and spatial domain. A new technology of information encryption and reconstruction is proposed combining with the theory of digital watermarking and iteration algorithm. This technology can overcome the limitation of phase detection errors. Furthermore, it can effectively hide and recover the embedded information, and the information can hardly be deciphered, which depends on the accuracy of phase information. Theoretical analysis and simulation results demonstrate that mean square error of the proposed algorithm can accomplish convergence by only several iterations within few seconds, which shows high robustness. It has a good performance in camouflage and fast-convergence, and could be widely applied in personal identity camouflage and information identity.