Image hiding technology plays an important role in the era of big data, especially in the fields of copyright protection and confidential communication. An image hiding method based on computational ghost imaging is proposed in this paper. The target image is encrypted by the computational ghost imaging system, random speckle is used as the system key, and the result of the binary ciphertext is embedded into the host image using the least significant bit algorithm. The corresponding watermark extraction rules and the second-order correlation algorithm are used to restore the target image. The simulation results show that the method has good concealment. The peak signal-to-noise ratio of the embedded watermark image can reach 50 dB. The authorized user can use the key to obtain the target image. At the same time, the proposed scheme has a certain fault tolerance. Image cropping is robust.