摘要
针对目前图像模糊检测算法复杂度过高和不能有效地实现图像局部模糊检测的问题,进行了图像局部模糊检测方法的研究,给出了模糊的数学模型,并基于此模型,提出一种简单有效的图像局部模糊的检测算法。该算法按照像素级搜索机制,对图像中各像素所在位置的相邻四邻域进行二维短时傅立叶变换,计算其频域系数的对数相关性,由此描述该区域的模糊程度,把它同预设阈值相比较,实现图像中局部模糊区域的检测。实验仿真显示,将该方法应用于篡改图像被动认证,可以有效地检测经过局部模糊和边缘模糊的篡改图像。
In order to reduce the complexity of conventional blur detection algorithms and efficiently detect the local-blur regions in the images, the paper gives a new blur detection method based on a study of local-blur detection and a machematical blur model. According to the 2-D Short-Time Fourier Transform, the method calculates the coefficients of four blocks around each pixel, the correlation value describing blur magnitude of this pixel＇ s area. Based on the comparison between this correlation value and the preset threshold, each pixel is divided into blurred class or un-blurred class. The proposed method is applied to the passive image authentication. The simulation results show the efficiency of this approach in revealing the local-blurred and the edge-blurred image tamper.
出处
《高技术通讯》
EI
CAS
CSCD
北大核心
2009年第7期718-723,共6页
Chinese High Technology Letters
基金
国家自然科学基金(60872114)
教育部博士点基金(20060280003)资助项目。
关键词
模糊检测
篡改检测
被动认证
blur detection, forgery detection, passive authentication
