摘要: 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。
关键词:
隐写检测,
扩频隐写,
小波奇异性分析,
高斯滤波器,
Fisher分类器
Abstract: 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.
Key words:
stego-detection,
spread spectrum steganography,
wavelet singularity analysis,
Gaussian filter,
Fisher classifier
中图分类号:
马 懿;张政保;冯 帆;王嘉祯. 基于小波奇异性分析的扩频隐写检测算法[J]. 计算机工程, 2009, 35(15): 159-161.
MA Yi; ZHANG Zheng-bao; FENG Fan; WANG Jia-zhen. Spread Spectrum Steganography Detection Algorithm Based on Wavelet Singularity Analysis[J]. Computer Engineering, 2009, 35(15): 159-161.