基于分割的空域图像隐写分析  

Steganalysis of Spatial Images Based on Segmentation

作　　者：汪然[1] 许漫坤[1] 平西建[1] 张涛[1] 

机构地区：[1]中国人民解放军信息工程大学信息系统工程学院,郑州450001

出　　处：《自动化学报》2014年第12期2936-2943,共8页Acta Automatica Sinica

基　　金：国家自然科学基金(61272490)资助~~

摘　　要：提出一种基于图像内容的空域隐写分析方法,该方法对图像进行分割,使分割得到的每一类子图像具有相同的统计特性,并对每一类子图像提取更加敏感的隐写分析特征,分别构造分类器进行训练和测试,由此对分割所得到的每一幅子图像都可以得到一个检测结果.对整幅图像的判决结果通过加权融合得到.实验结果表明,该方法具有良好的性能,尤其是针对自适应隐写方法,该算法的检测准确率提高更加明显.A new spatial image content based steganalysis algorithm is proposed. The given images are segmented to get the sub-images with similar statistical characteristics. More effective steganalysis features are extracted from each category of segmented sub-images respectively to build a classifier, Thus, a steganalysis result for each segmented sub- image can be obtained. The classifying results of all categories of sub-image are weighted fusing to obtain the overall steganalysis result. The performances shown by experiments demonstrate that the proposed algorithm can significantly improve the detection accuracy, especially for adaptive embedding methods.

关 键 词：信息隐藏 隐写分析 图像分割 纹理复杂度 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]