The improvement in reconstruction quality has always been the research focus of ghost imaging. A joint bilateral filter is embedded in the projected Landweber iterative algorithm, which is used for reconstruction in ghost imaging. This method can effectively remove the noises in the intermediate results of the projected Landweber iterative algorithm using the joint bilateral filter, and the reconstruction quality of the projected Landweber iterative algorithm is thus improved. The numerical simulation and the experimental results show that the proposed algorithm can effectively reconstruct the intensity image of the target. The proposed algorithm has higher reconstruction quality, if compared with the common ghost imaging reconstruction algorithms. At the same time, the influences of background noise and number of measurements on reconstruction quality are analyzed, and it is proved that the reduction of background noise and the increase of number of measurements improve the reconstruction quality of the proposed algorithm.