基于BP神经网络的隐写分析分类器设计

A Classifier of Steganalysis based on BP Neural Networks

设计并实现了基于BP神经网络的隐写分析分类器。首先对图像库中的图像进行格式变换,并使用扩展修改方向和钻石编码两种隐写方法进行不同嵌入率的隐写嵌入,然后计算载体图像和载密图像中平面域、DCT域和小波域的一些属性值作为特征。利用Matlab的模式识别工具箱搭建BP神经网络,用已知类别的图像特征训练分类器并进行分类测试。实验结果表明,多域综合特征可以实现良好的分类效果,能以较高的准确率识别出载体图像和载密图像。

A classifier of steganalysis based on back propagation neural network is designed and realized. Firstly,the images in image database are turned into gray-scale images, then some at ribute values of carrier images and stego-images are computed from the images＇spatial domain, DCT domain and wavelet domain. And the Matlab pat ern recognition toolbox is adopted to establish the BP neural？network classifier. The results show that the BP neural？network classifier based on three domains at ribute values performs wel in classifying.