As information technology develops rapidly, reversible data hiding in encrypted images (RDHEI) has drawn growing attention as its ability to safeguard the original image’s privacy while secretly storing the embedded data. Existing RDHEI schemes usually cannot meet the application scenarios where the content owner also need reversible data embedding. To tackle the problem, a new RDHEI scheme is proposed in this study to enable reversible data embedding for both the content owner and the data hider simultaneously. Specifically, the original image is first preprocessed by the content owner to generate two reconstructed images, each of the same dimensions as the original. Then, the content owner utilizes the proposed improved difference expansion (IDE) method for data embedding, and reserves room for subsequent data embedding. Finally, the images are stream ciphered and sent to two different data hiders. Data can be further embedded by each data hider in the received encrypted marked image to get the dual marked and encrypted image. When two dual marked and encrypted images are collected, the receiver can extract the data embedded by the data hider and the content owner separately, and recover the original image losslessly. Experimental results present that the proposed scheme can attain a relatively high embedding rate for textured images like ‘Baboon’, and in comparison with existing RDHEI schemes, the proposed scheme can allow the content owner to embed secret data in plaintext images, and can improve the embedding rate of secret data and enhance the security of the original image.