CAJ | 학술논문

本文提出以隐写分析为适应度函数的启发式蛙跳优化隐写通信模型，并以该模型设计了一种改进LSBM算法，即SFLA-LSBM算法。该算法以载体图像和载密图像的统计差异为优化目标函数，采用蛙跳算法优化隐写过程中的加减1序列，从而达到在隐写过程中最小化统计差异从而达到抵抗隐写分析的目的。实验表明，与传统的 LSBM 算法和现有改进LSBM算法相比，SFLA-LSBM有效提高了LSBM隐写安全性。SFLA-LSBM算法还具有在线学习功能，可抵抗新出现的隐写分析算法。
본문제출이은사분석위괄응도함수적계발식와도우화은사통신모형，병이해모형설계료일충개진LSBM산법，즉SFLA-LSBM산법。해산법이재체도상화재밀도상적통계차이위우화목표함수，채용와도산법우화은사과정중적가감1서렬，종이체도재은사과정중최소화통계차이종이체도저항은사분석적목적。실험표명，여전통적 LSBM 산법화현유개진LSBM산법상비，SFLA-LSBM유효제고료LSBM은사안전성。SFLA-LSBM산법환구유재선학습공능，가저항신출현적은사분석산법。
This paper proposes a heuristic shuffled frog leaping algorithm (SFLA) based steganographic model. An improved least significant bit matching (LSBM) image steganography ,abbreviated as SFLA-LSBM, is designed according to this new model. In SFLA-LSBM, an objective function is defined as the statistical difference between a cover image and its stego version. With the help of SFLA, an optimized plus-minus-one sequence, which is in fact the best feasible solution to the objective function, is found for each image . This sequence is embedded into the image to convey secret information as well as to minimize the change of image features. Compared with the classical LSBM steganography and other improved LSBM steganography alogrithms, SFLA-LSBM demonstrates better ability in resisting steganalytic algorithms. SFLA-LSBM is also adapted to resist newly presented steganalytic algorithms.