Image hiding is a technique for transmitting secret information under the cover of a digital image. It usually conceals sensitive information into images for the purpose of encryption. Currently, high embedding capacity and information security remain important research aspects of the image hiding. In this study, a secret image sharing scheme based on a reference matrix is proposed to enhance embedding capacity and verify data integrity. In the proposed scheme, a hill matrix is designed as a reference matrix and a location table is generated. Moreover, a location pair table is generated to ensure the uniqueness of data hiding locations. Then, leveraging the processing of the location pair table, as well as the mapping of the reference matrix and the location table, each pixel pair is exploited to conceal eight secret bits. Furthermore, based on the special construction of the hill matrix, a deception recognition mechanism is designed. This mechanism can detect deceptive behavior and identify tampered images by means of data hiding locations. The experimental results indicate that the proposed scheme achieves a higher embedding capacity and better deception recognition performance than that of most of existing schemes.