基于组合赋权及TOPSIS的隐写分析算法综合评估

分析了隐写分析技术在不同背景下的应用需求,提出了一种基于组合赋权及逼近理想解排序法(technique for order preference by similarity to ideal solution,TOPSIS)的隐写分析算法性能评估方法.该方法包含检测率、虚警率、可靠性、检测误差及算法运行速度等5个指标,先用熵权法确定指标权重,再根据层次分析法进行主观赋权,最后用TOPSIS实现对隐写分析算法的综合评估.实验结果表明,该方法可针对不同的性能指标要求选出最优的隐写分析算法,且对隐写分析算法性能的改进具有指导意义.

Comprehensive Performance Evaluation of Steganalysis Based on Combination Weight and TOPSIS

According to the different requirement of different applied fields of steganalysis,this paper proposes a steganalysis algorithm comprehensive evaluation method based on combination weight and TOPSIS.The method contains five performance indexes including true positive rate,false positive rate,reliability,detection error and detection speed.First,the entropy weight is put forward to give weights,then considering the defect of entropy weight and the estimator’s experience and intent,the AHP is used to give subjective weights,then the final weights are given by assembled weighting method.The evaluation and comparison to performance of steganalysis algorithm are implemented by using TOPSIS algorithm.Experimental results show that the evaluation algorithm can not only be applied to choose the best steganalysis method aim to different requirements of each index in different applications,but also can conduct to improve the steganalysis algorithm.