摘要
为了将基于传感器噪声的数字图像区域作伪检测技术扩展到更普遍的情况,即嫌疑图片同时受到剪切和缩放的情况,提出在作伪检测前,先对嫌疑图片的残留噪声进行预处理。预处理过程为:先在一定范围内用蛮力搜索寻找缩放因子;根据两个信号的标准互相关性最大值来估计剪切参数;利用估计出的参数对嫌疑图片的残留噪声进行反向操作,从而恢复图片噪声与参考相机(拍摄嫌疑图片所用的相机)指纹的同步性。实验结果显示,预处理后能成功检测出作伪区域。
In order to extend the image forgeries detection technology to a more general setting when the image under investigation has been simultaneously cropped and scaled.We proposed to make image preprocessing before detection.The scaling factor is found using a brute force search in a special range,and the cropping parameters are determined from the maximum of the normalized between two signals.At last the noise residual from the suspicious image is reverse-operated using the estimated parameters to recover the synchronization between the image noise and the camera fingerprint.The experiment results show that the tamper area can be detected successfully after preprocessing.
出处
《贵州大学学报（自然科学版）》
2011年第2期74-77,共4页
Journal of Guizhou University:Natural Sciences
关键词
相机指纹
数字取证
复制-粘贴篡改
作伪检测
camera fingerprint
digital forensics
copy-past tamper
forgeries detection
