图像重采样检测  

Tampered-Image Detection

作　　者：耿振民[1] 朱丹[2] 

机构地区：[1]上海华御信息技术有限公司,上海200433 [2]江南大学物联网工程学院,江苏无锡214122

出　　处：《信息安全与通信保密》2013年第9期55-59,共5页Information Security and Communications Privacy

摘　　要：图像的重采样篡改检测和JPEG双重压缩的篡改检测都是图像篡改取证的重要方法。大多数重采样篡改检测都会忽略JPEG压缩对图像的影响,而文中提出的方法主要是利用图像经过重采样和JPEG压缩之后的周期性、图像的自然统计特征对重采样图像进行检测,首先提取出可以区分原图像和重采样图像的重采样因子和JPEG压缩因子,然后提取图像的4个统计特征,最后将6个因子组合成特征向量,利用支持向量机对其进行训练和分类。实验证明,该算法对于缩放因子小于1的图像和原图像都有较好的检测率。Image-resampling tamper detection and JPEG double compression tamper detection are both the methods for image tamper detection. Most resampling tamper detections ignore the influence of JPEG compression. This paper proposes that the periodicities of resampling, JPEG compression and characteristics of natural images could be utilized to detect the images different from the normal images. First, the factors of resampling and JPEG compression that could distinguish the original image from the sampling image are extracted; then the four statistical characteristics are picked up from the existing images ; and finally the six factors are combined into the feature vectors so that support vector machine could be used for training. Experiment shows that the proposed idea has a fairly good efficiency in the detection of original image and resampling image with a scaling factor of less than 1.

关 键 词：图像取证 重采样 JPEG压缩 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]