融合统计矩和共生矩阵特征的隐写分析算法

作者：化晓茜; 胡永健; 马赛兰; 刘琲贝

来源：中国,陕西省,西安市.

• 图像处理
• 隐写分析算法
• LSB匹配
• 纹理特征
• 统计矩
• 差分灰度共生矩阵

　　The steganographic embedding often alters the pixel values of cover image, and thus changes image textures.If we view the embedded message as a kind of texture, steganalysis can be thought of as the problem of image classification based on image textures.For the detection of LSB matching, we propose a new prediction-error function, which can better emphasize the image residual.We then calculate both macro and micro texture characteristics.The macro texture characteristics include two descriptions from statistical moments.They are respectively fused with the micro texture characteristic derived from the gray difference co-occurrence matrix.The SVM (support vector machine) is used for image classification.We have tested our two methods on five commonly used image databases.Experimental results show that they have good performance in detecting the embedding by LSB matching and they are better than other two similar methods.

• 单位
  兰州交通大学; 华南理工大学