针对隐写分析冗余评估指标的属性约简算法

针对隐写分析评估指标体系中存在的知识冗余，提出了一种隐写分析评估指标的属性约简方法。在分析存在知识冗余的指标体系的基础上，采用BCC算法求出各评估指标初始区间及其约简，并用之将仿真实验数据离散化；之后将离散化矩阵作为HORAFA算法的输入，输出即得到最终的属性约简。仿真实验验证了该属性约简方法对冗余评估指标约简的有效性，通过对属性约简前后的评估算法兼容度分析，表明属性约简使评估结果的可靠性显著提高。

Attribute Reduction for Steganalysis Redundant Indicators

For the existence of knowledge redundancy in steganalysis evaluation indicators, an attribute reduction for steganalysis indicators is proposed. Based on analysing the indicators knowledge redundant, the algorithm uses BCC algorithm to get initial interzone and its reduction of each index, with which data of experiment result are despersed.The discrete matrix is used as the input of Heuristics Optimal Finding Reduct Algorithm based on Frequencies Attributes(HORAFA) algorithm,and then the final reduction result of steganalysis evaluation indicators is obtained. The experiments results show that the algorithm is effective in steganalysis evaluation indicators reduction. It is proved that the attribute reduction algorithm can increase the reliability of evaluation results by analysing the compatibility of given evaluation algorithm before and after attribute reduction.