图像模糊检测方法

针对数字伪造图像中最常用的模糊操作手段之一的高斯模糊操作,文中提出了一种基于同态滤波和DCT功率谱的图像模糊检测方法.该方法分析了离焦模糊和人工模糊的边界特点,首先利用同态滤波增强人工模糊操作的边缘,并提取增强后图像的二值边缘,从而区分自然模糊和人工模糊,以减小实验中出现的误判,再利用图像的DCT功率谱的分布检测出模糊篡改的区域.该方法能够有效地检测出图像的模糊操作,并进行准确定位.

Image Blur Detection

As Gaussian blur operation is the most commonly-used means in digital image forgery, this paper proposes an image blur detection method based on homomorphic filtering and DCT power spectrum. The method analyzes the boundary characteristics of defocal blur and artificial fuzzy, enhances the edge of artificial blur used by the homomorphic filtering, and extracts the binary edge of the enhanced image, thus makes distinctions of between natural blur and artificial blur and reduces false positives, and then uses the distribution of DCT power spectrum to detect the fuzzy tampered region. Experiment indicates that this method can effectively detect the blur operation and implement accurate positioning.