基于Boosting算法融合的图像隐写分析方法

现有盲检测技术在实际检测中，由于嵌入算法未知导致检测困难。为此，提出一种基于Boosting算法融合的图像隐写分析方法。通过训练分类器建立不同隐写算法下的分类器模型，利用Boosting算法计算各分类器的分类性能，对各分类器的概率输出进行融合，得到最终检测结果。基于典型空间域隐写算法和JPEG隐写算法的实验结果表明，该方法实现了对多种隐写算法的有效检测，应用Boosting算法融合后整体检测性能提升了约2％。

Image Steganalysis Method Based on Boosting Algorithm Fusion

The existing detection algorithms are difficult to obtain high detection accuracy when applied to the condition, in which the embedding algorithm of the stego-images is unknown. ＇Therefore, this paper proposes a steganography-unknown image steganalysis method based on Boosting fusion. It obtains various classifying results by establishing steganography algorithm classifier models in the training phase, and acquires the performance of these classifies according to the Boosting algorithm. The final detection result is obtained by combinational rule based on probability output. The detection work is presented to attack the current different spatial domain and JPEG steganographic algorithms. Extensive experimental restdts show that this proposed method is effective for multi-steganographic algorithms, and Boosting takes advantage of the individual strengths from each detection system and whole detection performance is probably increased by 2%.