Abstract:

A method to discriminate stego and noise image is proposed. Considering gaussian white noise and salt and pepper noise, which exist in images frequently, the histogram characteristic function center of mass is used to classify original and stego and noise images based on additive noise modelable. The variance of high frequency subband’s coefficients in wavelet domain is used to discriminate stego and noise image. Experiments on a wide variety of images indicate the veracity of this method.

Key words: image steganography, noise image, histogram characteristic function, wavelet decomposition

摘要：

提出一种辨识图像隐写与自然噪声的方法。从图像中常见的2类噪声(高斯白噪声和椒盐噪声)出发，基于加性噪声模型，利用图像直方图特征函数质点区分原始图像和隐写噪声图像，利用小波高频子带系数差分方差识别隐写图像与噪声图像。对大量隐写和噪声图像进行实验，结果表明，该方法可有效辨识图像隐写和噪声。

关键词: 图像隐写, 噪声图像, 直方图特征函数, 小波分解

CLC Number: 

• N945