A Systematic Review of RDHEI with Consistent Experimental Evaluation

Reversible data hiding in encrypted image (RDHEI) has recently seen high attention due to its superior embedding capacity (EC) and preservation of image quality. Central to their efficacy is the accuracy of the prediction techniques employed, coupled with the method for systematizing and compressing the obtained predicted errors. This paper delves into an extensive analysis of various predictors in high-capacity RDHEI (HC-RDHEI), covering a spectrum from basic approaches like difference prediction, median edge detection and gradient-based predictors to more advanced ones such as quad-tree-based prediction. It also meticulously evaluates the systematization and compression strategies for errors across these prediction methods. Using two datasets for experimental validation, this research provides a holistic overview of HC-RDHEI techniques and their prediction paradigms. This in-depth assessment aims to help future researchers and industry professionals, enabling well-informed decisions regarding prediction method selection in HC-RDHEI applications.