Local Binary Pattern Based Robust Coverless Information Hiding Approach

To address the drawbacks of existing secret data hiding techniques, such as their high computational demands for training deep neural networks or their limited hiding capacity, we present a new approach. Our proposed approach utilizes Local Binary Pattern codes to generate hash sequences for concealing secret data, offering a more efficient solution. These hash sequences are used to encode the secret information within the cover images. To enhance security, we encrypt the mapping file using a Paillier cryptosystem. The recipient can successfully extract the secret information once the mapping file is decrypted and reverse the hash sequence generation algorithm is applied. The effectiveness of the method is demonstrated through extensive experimentation and comparison with the existing approaches. Results indicate that the proposed method not only reduces the computational cost but also outperforms traditional methods in terms of security and robustness. Our method reports approximately 95% accuracy in data recovery after common image processing attacks, such as scaling, noise addition, and image filtering, and allows to hide 72 bits of secret data per image.