imaging modalities, computational ghost imaging (CGI) and visual cryptography (VC), [5,6] have demonstrated their great potential in optical encryption. [7][8][9][10][11][12] In CGI, a sequence of computer-generated mask patterns are projected onto the target object, and a single-pixel detector is employed to record the relevant intensity values reflected or transmitted from the object. A decryption-like computational process is then applied to reconstruct the object image from the mask patterns (i.e., keys) and intensity values (i.e., ciphertext). Various optical encryption schemes based on CGI have been proposed for multipositioned or multicolored image encryption. [13,14] Different from the computational reconstruction of CGI, VC enables a direct observation of the secret image by naked eyes. In VC, a secret image is split into multiple shared key images. The predesigned shared images with incoherent overlays can decode the secret image visually by gathering their independent subinformation. By combining holography and VC, optical encryption schemes have been developed by hiding the secret image into different pieces of phase keys. [15][16][17][18] Meanwhile, rapid progress has been made in optical encryption, thanks to the development of metasurfaces. [19][20][21][22][23][24][25][26] Metasurfaces, consisting of single-layer or fewer-layer spatially variant meta-units, enable unprecedented control over the phase profile and other degrees of freedom of light with subwavelength resolution. [27][28][29][30][31][32][33][34][35][36][37][38][39][40][41][42] With polarization or wavelength control, optical Exploiting various degrees of freedom of light, metasurfaces have unique advantages in multiple-channel information storage and demonstration, which thereby provides a novel platform to convey the keys and cipher images for different encryptions. Following the secret sharing principle of visual cryptography (VC), the authors here successfully embed both the keys and cipher images of computational ghost imaging (CGI) encryption into the holographic metasurface-images (meta-images). The decryption process starts with key retrieval via optical observation of overlapped meta-images, followed by a compressive CGI calculation to reconstruct the target images according to the obtained key and steganographic cipher images with a high compression ratio of 4. By integrating metasurface imaging, VC, and CGI, the authors' proposed encryption scheme exempts conventional key distribution and transmission of CGI, enhances the security by secret sharing of VC, and increases the amount of hiding data contained in meta-images with compressive sensing.