In order to improve the practicability of compressed sensing ghost imaging and solve the problem of associative imaging failure caused by the loss of sampled data and the inability to repeat sampling in the scene, a expansion method of packet loss data in ghost imaging is proposed. First, the influences of different packet loss data on imaging performance are analyzed. Then, the image quality is improved by grouping the sample data and extending the sample results with missing phenomena. The simulation and experimental results show that, compared with the traditional method, the grouping expansion method can reduce the influence of packet loss data on the imaging quality, which is beneficial to further promote the practical application of ghost imaging.