Reconstruction the computer generated Fresnel hologram of complex 3D object based on compressive sensing (CS) is presented. The hologram is synthesized from a color image and the depth map of the 3D object. With the depth map, the intensity of the color image can be divided into multiple slices, which satisfy the condition of the sparsity of CS. Thus, the hologram can be reconstructed at different distances with corresponding scene focused using the CS method. The quality of the recovered images can be greatly improved compared with that from the back-propagation method. What’s more, with the sub-sampled hologram, the image can be ideally reconstructed by the CS method, which can reduce the data-rate for transmission or storage.