拼接篡改图像的色温估计取证方法

An Authentification Method for Splicing Images with Color Temperature Estimation

针对近年来图像伪造现象严重,而传统的图像取证方法过于依赖人工经验,缺乏严谨的取证依据的问题,提出一种基于色温估计的拼接篡改图像取证方法.利用改进的平均色差计算方法对图像进行色温估计;然后根据色温估计结果计算关键区域与参考区域之间的色温距离;最后与实验获取的色温距离阈值进行比较,判断图像中是否存在拼接篡改操作所引入的色温不一致现象.实验结果表明,该方法能够为图像取证工作提供科学量化的参考依据.

Image forgery has been a serious issue in recent years. Traditional authentification methods depend on experience excessively, and lack strict and consistent standards. In this paper, a splicing image authentification method based on color temperature estimation is proposed. Our method uses an improved chromatic aberration calculation algorithm to estimate color temperature of key areas in images. Then distances between these regions are computed with the results of color temperatures. After the above procedure, the distance is compared with a threshold value based on experiments to find whether color temperatures are inconsistent among these images. Experiments demonstrated that our method can verify splicing images exactly, and provide scientific evidences for image authentification.