基于质量评价量和方差分析的图像篡改检测

摘要

Aiming at the problem of digital image slicing, this paper proposes a detection method based on image qualities and analysis of variance. Four kinds of noise are used to simulated the tampering of image, and analysis of variance is used to select the image quality measures which are more sensitive to image splicing. Combined with the characteristic function moments of wavelet sub-bands, all features from given image and its predicted error image are extracted. LibSVM is adopted as the classifier to train and test the given image. Simulation results show that average detection accuracy of the proposed method is increased by 1.3％～ 15％ compared with other methods.%针对数字图像篡改问题,提出一种利用方差分析选择图像质量参数以实现篡改图像盲检测的方法.使用4种噪声模拟数字图像篡改方式,通过方差分析选出对篡改较为敏感的图像质量评价参数,结合图像小波高阶特征函数统计矩,建立篡改检测算法的特征模型,利用LibSVM对其进行训练和分类.仿真结果显示,该方法相比同类方法平均检测率提高了1.3%-15%.