基于训练集构造的图像通用盲检测算法改进

Improvement of Image Universal Blind Detection based on Training Set Construction

现有通用盲检测技术普遍存在泛化问题,导致检测器实用性大大下降。根据正交设计原则构建隐写率失匹配集合,隐写算法失匹配集合和图像源失匹配集合,分别分析检测SPAM分析算法和Rich Model分析算法在隐写率失匹配,隐写算法失匹配和图像源失匹配方面的检测率。并根据测试结果提出通过训练小隐写率图像集,训练多类隐写算法,图像预分类和改进IQM分析算法几种方案解决泛化问题,实验结果显示经过改进后隐写分析算法性能得到明显提升。

The practicability of existing universal blind detection reduced greatly due to the generalization problem. Ac- cording to the principle of orthogonal design, this paper builds three sample sets of embedding rates mismatch, embed- ding algorithms mismatch and image sources mismatch between the training sample and the testing sample. The three sets are used to test the detection error rates of SPAM and Rich Model in the case of embedding rates mismatch, embed- ding algorithm mismatch and image source mismatch. This paper proposed several methods to improve the generalization ability of the universal blind detection, including training the sample by small embedding rates, learning various kinds of embeddingalgorithms, pre-elassifying the testing sample and improving the IQM algorithm. The results show that the that the performance of the improved algorithm is significantly improved.