Chang and Hwang proposed a delicate visual cryptography scheme for generating meaningful transparencies. However, in their method, the size of each transparency is nine-fold of the secret image. For saving the storage of the transparencies, a non-expansible visual cryptography scheme for generating meaningful transparencies is proposed. There are some characteristics in the proposed method: (1) each transparency and the secret image are with the same size; (2) each transparency is meaningful; (3) the secret can be decrypted by human eyes; (4) the contrast the stacking result is a little higher than Chang and Hwang’s method.