An experimental setup of ghost imaging via sparsity constraints (GISC) for a rough surface target is designed, and the influences of both the numerical aperture of receiving system and rough target size on imaging performance of GISC are investigated and analyzed by using this setup. The results demonstrate that the imaging quality of GISC for the rough surface target is positive in relation to both parameters. This research could be considered as an instructive role for the optical system design of a GISC system.