(1)Propose a more flexible side information selection approach with less constraints for ‘‘multilevel embedding’’ based RDH framework without those unnecessary constraints(2)Offer a pattern based rapid performance evaluation method for multilevel embedding process to reduce compu- tation complexity;(3)Design a novel problem-oriented evolutionary algo- rithm, i.e., Transfer Learning based Genetic Algorithm (GA), to achieve high-efficiency search of optimized side informa- tion within an entire huge solution space.