Reversible Data Hiding Algorithm Based on Adaptive Predictor and Non-uniform Payload Allocation

Reversible data hiding is a technique that enables the secure embedding and complete extraction of data without reducing the quality of the carrier image. It has significant application value in fields such as medical images, military images, and digital forensics. However, existing reversible data hiding methods often need clarification on embedding capacity, image quality, and the trade-off between computational complexity and robustness. This paper proposes a reversible data hiding algorithm based on adaptive predictor and non-uniform payload allocation. The algorithm first uses an adaptive predictor to predict the image and then dynamically allocates different embedding bits according to the size and distribution of the prediction error, thus achieving non-uniform payload allocation. The algorithm only changes the low bits of the prediction error when embedding data, thus ensuring the high fidelity of the image quality. The algorithm can fully recover the original image when extracting data, thus achieving reversibility. The paper conducts experiments on various types of images, and the results show that the algorithm outperforms existing reversible data hiding methods in terms of embedding capacity and image quality while having lower computational complexity and stronger robustness.