一种基于自适应闭值的图像伪造检测算法

图像伪造检测是数字取证领域一个发展迅速的研究方向。复制一移动是最常见的图像伪造方式之一，其目的是通过隐藏或克隆对象来创建新的图像内容场景。复制一移动伪造检测的主要依据是图像中存在较大面积的相同或非常相似的区域对。针对以往检测方法对图像中存在同质纹理或均匀区域检测困难以及相关参数阂值选择不确定等现状，提出一种基于自适应阂值的图像复制一移动伪造检测算法，该算法不但使相关阂值的选择和估计更合理，而且能够自动识别和定位伪造区域。通过在包含同质或均匀区域的彩色伪造图像中的实验，进一步验证了本算法的有效性。

Adaptive Threshold-based Detection Algorithm for Image Copy-move Forgery

Image forgery detection is a burgeoning research field in digital forensics. As a most common way of image tampering, copy-move forgery is used to conceal objects or clone regions to produce a non-existing image scene. The target of copy-move forgery detection is to identify bigger duplicated image regions which are same or extremely similar. We reviewed several methods proposed to achieve this goal. These methods failed in detecting digital images with homogeneous texture or uniform areas and selecting the appropriate thresholds. We presented a novel adaptive thresholdbased detection algorithm for image copy-move forgery,which might be applied to the color images with homogeneous or smooth regions and identified and located forged image regions automatically if only reasonable thresholds were estimated. Experimental results on several forged images with various homogeneous or uniform regions were presented to demonstrate the effectiveness of the proposed algorithm.