基于小波奇异性分析的扩频隐写检测算法

Spread Spectrum Steganography Detection Algorithm Based on Wavelet Singularity Analysis

针对扩频隐写会破坏图像局部平稳性而使其产生高频奇异性的缺点,提出一种基于小波奇异性分析的扩频隐写检测算法。通过分析待测图像不同尺度下小波系数模极大值数量的变化情况,提取8维特征向量作为Fisher分类器的输入向量并对其进行训练。对测试样本的检测和攻击实验结果表明,该算法的平均检测率达到80%以上,能够检测出隐写的大致频带范围并实施有效的滤波攻击,为隐秘信息的进一步提取奠定了基础。

Aiming at the shortcoming that spread spectrum steganography breaks image local stationarity, this paper presents a detection algorithm for spread spectrum steganography based on wavelet singularity analysis. It extracts 8 dimension feature vector as the input vector of Fisher classifier by analyzing the changes of wavelet coefficients modulus maximum on different scales of images to be detected, and uses a mass of samples to train Fisher classifier. The detection and attacking experimental results prove that the average detection rate of the algorithm is more than 80%, and it can detect the spectrum range in which the secret message is hidden and implement effective attack, which lays the foundation for extracting the stego message.