中文摘要：

分析了JPEG图像的信息隐藏方法及各种攻击算法，提出了一种新的基于支持向量机（SVM）的目检测算法。通过分析研究与信息隐藏相关的多个图像统计特征量，设计了图像DCT系数的直方图、边界不连续性、相关差值、相关直方图等值组成特征向量，然后采用SVM方法进行分类，实现了一种新的图像隐密分析方法。实验结果表明，这是一种有效的、高精度的盲检测方法，能够准确识别出图像是否隐藏有秘密信息。

英文摘要：

After introducing some information hiding methods and steganalysis techniques for JPEG images, this paper proposes a new SVM-based blind steganalysis algorithm for JPEG images, The algorithm analyzes some statistical features related to information hiding, designs the features about the image＇s histogram, the discontinuities of the block boundaries, correlative variance and correlative histogram to compose the feature vector, and uses this vector to classify stego-images by the SVM method. The experiments show that the new algorithm is an effective blind steganalysis method with quite accurate detection.