摘要
MME(Modified Matrix Encoding)作为一种新兴的JPEG图像隐写技术,具有隐写量大、对图片的改动小、抗检测性优于传统隐写算法等优点。文中在采用Markov特征的基础上,加入了能够更好描述图像局部纹理的局部二值模式(LBP)与局部顺序对比模式(LOCP)特征,并应用Ensemble分类器完成分类和识别。通过对UCID图像库的大量实验,得到了一种针对MME算法的最优特征组合。相比传统的隐写分析方法,文中所提出的方法具有更好的检测正确率。
As one of the emerging JPEG image steganography, MME（Modified Matrix Encoding） has the advantages of large embedding capacity, small change of the image, better resistance to detection than traditional steganography algorithm, etc. Based on the Markov features, the LBP and LOCP features are proposed in this paper, which could better describe local texture of image. The Ensemble classifier should be adopted for classification. Through a large number of experiments on UCID image database, the optimal feature combination for MME algorithm could be achieved. Comparing with the traditional steganalysis methods, the proposed method in this paper has better detection accuracy.
出处
《信息安全与通信保密》
2014年第4期103-109,共7页
Information Security and Communications Privacy
基金
国家自然科学基金资助项目--唇语特征在身份认证中若干关键技术研究(批准号:61271319)
