小波统计量和多类支持向量机的密写检测算法  

Detection of Steganography Based on Wavelet Statistics and Multi-Class SVM

作　　者：刘佳[1] 杨晓元[1] 吴艺杰[1] 

机构地区：[1]武警工程学院电子技术系网络与信息安全武警部队重点实验室,西安710086

出　　处：《计算机工程与应用》2006年第26期105-108,共4页Computer Engineering and Applications

基　　金：国家自然科学基金资助项目(编号:60473029);信息安全教育部重点实验室课题资助项目(编号:200409)

摘　　要：提出了一种基于小波统计量和多类支持向量机的彩色图像密写检测算法。为克服以往将彩色图像转化为灰度图像引起的各颜色通道相关性损失的不足,算法建立了彩色图像统计模型。对彩色图像每个颜色通道分别进行小波分解,根据小波分解系数绝对值和绝对值线性预测的对数误差生成特征向量,并采用多类支持向量机进行模式分类。在特定嵌入率下对几种常见的密写软件生成的密写图像进行测试。实验表明此算法具有一定的通用性,对密写图像具有较高的识别率。A new detection of steganography algorithm based on wavelet statistics of color images and multi-class SVM is proposed.In order to capture some regularities which are ignored when converting color images into grayscale images among the color channels,statistical models for color images are built.The wavelet decomposition is implemented in each color channel,the magnitude of decomposition coefficients and the log error between the actual coefficient and the predicted coefficient magnitudes are used to yield statistics.The multi-class support vector machine algorithm has been employed in the pattern discrimination,Stego images created by several tools are tested under certain embed rate.The experimental results show that the algorithm has stronger universal performance and higher discriminating rate.

关 键 词：密写图像检测 多类支持向量机 小波分解 核函数 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]