Abstract: In recent years, with the advent of all kinds of advanced image processing algorithms and the corresponding image processing hardware and software, digital image has become much easy to forge even for non-professional users, and the forged image is hard or even impossible to be distinguished from the original image by naked eyes. This paper concentrates on a specific category of digital image forgery known as copy-move forgery, and proposes a method to automatically detect and localize duplicated regions in digital images. An image is tiled with overlapping blocks and each block is characterized by blur invariants, which is calculated from blur moment invariants. It determines the image tamper regions in terms of predetermined similarity criteria. The experimental results show that the method has obvious advantage in anti-blur processing in contrast with the method based on PCA（Principal Component Analysis）.

Key words: image forensics, image tempering, copy-move forgery, blur moment invariants

摘要： 随着各种高级图像处理算法以及相应图像处理软硬件的出现，即使非专业人士也很容易篡改图像，并使人肉眼很难甚至无法识别。针对一种常见的图像篡改——复制粘贴伪造，提出了一个能自动检测并标识数字图像中复制区域的方法。将图像分成多个重叠块，每块的特征用由模糊不变矩计算得到的模糊不变量表示，按照预定的相似标准来确定图像篡改区域。实验结果表明：相对于基于PCA的方法，该方法在抗模糊处理方面具有明显的优势。

关键词: 图像取证, 图像篡改, 复制粘贴伪造, 模糊不变矩