Computational ghost imaging (CGI) can easily penetrate a static scattering medium. However, when the scattering medium changes dynamically, the measured light intensity is subject to nonlinear changes caused by the medium; as a result, the light intensity of each measurement exhibits a drift, and the reconstructed image contains lot of noise due to the interference. To improve imaging results penetrating a dynamic scattering medium, this study proposes a method to correct the measured value of each point. In this method, patterns with the reflectivity of 1 are added to a measurement matrix. The attenuation factor is obtained at each time by recording the change of light intensity after projecting the pattern, and the compensation coefficient at the middle time is calculated. According to the compensation coefficient, the light intensity distortion caused by the change of the medium’s attenuation factor is corrected for each point, and finally, the transformed values are restored and calculated. The reconstruction results show that the proposed method can effectively deal with the aforementioned noise problem of the CGI technology encountered while penetrating a dynamic scattering medium.