摘要
设计并开发了一种基于DCT系数能量及能量差特征,三层前馈非线性神经网络来作为分类器的JPEG图像隐藏信息的盲检测系统。整个系统的开发环境是VC++6.0,采用了多线程XML技术以及SQL Server数据库,实现了JPEG图片的批量盲检测,检测信息存储以及报表生成。系统对三千多幅JPEG图像进行盲检测测试,对于多种隐写方法取得平均检测率90.085%,包括低嵌入率的JPEG含秘图像和平滑、锐化、缩小、剪切和再压缩处理后的JPEG净图。
A blind detecting JPEG Steganographic Content system was designed and developed whose class is based on the characteristics of DCT coefficients energy and 3-layer forward neural network. The whole system was developed in VC + + 6.0, using the multi-thread, XML , SQL Server database and the JPEG library of Thomas G. Lane, and a blind detecting JPEG Steganographic Content system was realized which can store detecting information and create reports. This system has detected more than 3000 pictures and the average detecting rate of various Steganographic methods is 90. 085%, including jpeg pictures with low stega rate and cover pictures which are smoothened, sharpened, shirked, cuted or re-compressed.
出处
《计算机应用》
CSCD
北大核心
2007年第6期1527-1530,共4页
journal of Computer Applications