非对称特征图像区域伪造准确检测方法仿真

摘要

Accurate detection of forged regions of asymmetric feature image can effectively improve the realness of digital image.In traditional way,the elimination of mismatch of image transformation parameters is ignored,resulting in low detection accuracy.In this paper,we focus on a forgery and detection method of asymmetric feature image based on Fourier transform.We used Fourier Mellin transform to extract block feature of asymmetric characteristic image,and then used the random sample consistency algorithm to estimate affime transformation parameters between characteristic matching pairs in asymmetric characteristic image region and eliminate the mismatch.According to regional association diagram of asymmetric characteristics image,we determined the complete forged region of asymmetric feature image.Experimental results prove that the proposed method can effectively locate the forged area,which has strong robustness for the rotation of copy area and processing operation after brightness change.%对非对称特征图像的伪造区域进行准确检测,能够有效提高数字图像的真实性.对图像区域伪造的准确检测,需要采用傅里叶-梅林变换提取图像块特征,对图像仿射变换参数进行估计并消除错配,完成图像伪造区域的检测.传统方法对目标图像进行分块,依据提取图像子块获取局部网点密度特征,但忽略了对图像变换参数错搭的消除,导致检测精度偏低.提出基于傅里叶变换的非对称特征图像伪造检测方法.采用傅里叶-梅林变换提取非对称特征图像块特征,采用随机样本一致性算法对非对称特征图像区域特征匹配对之间的仿射变换参数进行估计并消除错配,依据构建非对称特征图像区域关联图确定完整的非对称特征图像伪造区域.实验结果表明,所提方法可有效定位出伪造区域,对于复制区域的旋转、亮度变化后处理操作具有较强的鲁棒性.