摘要
图像的局部复制粘贴篡改技术,是最常见的一种图像伪造方式,对此提出一种基于小波矩的图像复制粘贴篡改检测算法.首先通过变分水平集活动轮廓模型初步确定图像篡改的可疑区域;然后对每一块可疑区域利用小波矩算法提取其小波矩特征;接着利用余弦相关性测度判别可疑区域的相似性;最后定位图像的篡改区域.实验结果表明本算法能够有效提取可疑区域,并进一步定位篡改区域.此外,算法对图像前景篡改区域的平移、旋转和缩放具有较强的鲁棒性.
The regional copy-move tampering for a digital image is one of the most common forgery methods. This paper proposes a detection algorithm based on wavelet moment to detect the location of image copy-move tampering. Firstly, we tentatively figure out the suspicious tampered areas in an image by the active contour model based on variational level sets. Secondly, the wavelet moment algorithm is employed to extract wavelet moment features from all suspicious areas. Finally, the real tampered areas are located by using cosine correlation measure to judge the similarity of those suspicious tampered areas. The experiment demonstrates the effectiveness of above mentioned algorithm and its good robustness to the operations of translation, rotation and scale for foreground tampered areas.
出处
《计算机系统应用》
2013年第12期108-112,共5页
Computer Systems & Applications
基金
福建省自然科学基金(2011J01346)
福州大学科研基金(022336)
关键词
图像盲取证
小波矩
图像复制粘贴篡改
活动轮廓模型
余弦相关性测度
blind image forensics
wavelet moment
image copy-move tampering
active contour model
cosinecorrelation measure
