作者：吴冰洁, 杜振龙

作者单位：南京工业大学电子与信息工程学院, 江苏 南京 211800

关键词：图像伪造组件检测;基元树;包络盒图像;组件拆分;对称因子差异;密集度差异

摘要：

为解决当前图像伪造检测方法因采用逐像素或逐块比对的鉴别方式而导致耗时长、内存空间消耗大以及伪造定位准确度较低等不足,提出基于基元树的建筑物图像伪造组件检测算法。该算法根据建筑物的对称性特征,提取目标图像的基元,获取包络盒图像,再以其为根结点,借助分割与嵌入技术,对组件进行拆分,构造出基元树;再利用基元树来表征建筑物图像,凸显基元之间的空间分布关系,并定义非聚合组件的对称因子差异以及聚合组件的密集度差异计算模型,通过检索基元树结点对应的非聚合组件对称因子差异值或聚合组件密集度差异值,快速检测出建筑物图像中的疑似伪造组件,即疑似伪造图像区域。实验结果表明:与当前图像伪造检测技术相比,所提方法的检测速度更快,约19.92 s;且准确度更高。所提基于基元树的建筑物图像伪造组件检测算法能够高效准确检测出建筑物图像伪造区域。

Facade forgery component detection algorithm based on element tree

WU Bingjie, DU Zhenlong

College of Electronics and Information Engineering, Nanjing University of Technology, Nanjing 211800, China

Abstract: An efficient forgery image-detecting approach based on element tree was proposed in the paper to solve the time-and big space-consuming problems and low-location accuracy in current forgery detection algorithms namely pixel-to-pixel or block-to-block matching. First, box-bounding images were acquired by extracting the elements of the input building image according to the symmetry of the building. Second, the images were used as a root node to create an element tree by splitting the forged image components in the way of segmentation and embedding. Third, the spatial distribution relationship between the elements was highlighted by using the element tree to characterize the building image. Meantime, two computing models were defined for the symmetrical factor differences of non-polymeric components and the intensity differences of the polymeric components. The suspected forged components of the building image were rapidly detected by retrieving the above-mentioned differences corresponding to the roots of element tree. The experiments demonstrate that the proposed algorithm, with a detection speed of 19.92 s, is more efficient and accurate than the existing algorithms in regard to the identification of forged building images.

Keywords: images forgery components detection;element tree;box-bounding images;component split;symmetry factor differences;intensity difference

2015, 41(11): 93-100  收稿日期: 2015-02-05;收到修改稿日期: 2015-04-19

基金项目: 国家自然科学基金项目(61073098);江苏省六大人才高峰基金项目(2012-WLW-023);教育部高等学校博士点基金项目(20113221120003)

作者简介: 吴冰洁(1990-),女,江苏南京市人,硕士研究生,专业方向为图像伪造检测、信息安全。

参考文献

[1] 李满满,杜振龙,沈钢纲. 基于奇异值加权的图像复制粘贴伪造盲检测[J]. 计算机工程与设计,2012,32(12):4125-4129.
[2] Shivakumar B L, Baboo L D S S. Detection of region duplication forgery in digital images using SURF[J]. IJCSI International Journal of Computer Science,2011,8(4):199-205.
[3] 杜振龙,杨凡,李晓丽. 利用SIFT特征的非对称匹配图像拼接检测[J]. 中国图象图形学报,2013,18(4):442-449.
[4] Amerini I, Ballan L, Caldelli R. A SIFT-based forensic method for copy-move attack detection and transformation on recovery[J]. IEEE Trans on Information Forensics and Security,2011,6(3):1099-1110.
[5] Cao Y, Gao T, Fan L. A robust detection algorithm for region duplication in digital images[J]. International Journal of Digital Content Technology and its Applications,
2011,5(6):95-103.
[6] Kang X, Wei S. Identifying tampered regions using singular value decomposition in digital image forensics[C]//Proceedings of International Conference on Computer Science and Software Engineering. Washington DC:IEEE Compute Society,2012,35(13):926-930.
[7] Huang Y, Lu W, Sun W. Improved DCT-based detection of copy-move forgery in images[J]. Forensic Science International,2011,206(13):178-184.
[8] Martinovic A. A three-layered approach to facade parsing[J]. Computer Vision-ECCV,2012,32(8):416-429.
[9] Teboul O. Segmentation of building facades using proced
ural shape priors[C]//Proceedings of IEEE Conference on Computer Vision and Pattern Recognition(CVPR),San Diego,2010,12(11):3105-3112.
[10] Fan L B, Musialski P, Liu L G. Structure completion for gird layouts[J]. ACM Transactions on Graphics,2014,
33(6):1-10.
[11] Zhang H, Xu K, Jiang. Layered analysis of irregular facades via symmetry maximization[J]. ACM Transactions on Graphics,2013,32(4):1-10.
[12] Hu S M, Zhang F, Wang M, et al. a patch-based image representation for interactive library-driven image editing[J]. ACM Transactions on Graphics,2013,32(6):2504-2507.
[13] Lin J, Cohen O D, Zhang H, et al. Structure-preserving retargeting of irregular 3D architecture[J]. Journal of Computer-aided Design & Computer Graphics,2012,30(6):61-64.
[14] Bay H, Ess A, Tuytelaars T. SURF:speed up robust features[J]. Computer Vision and Image Understanding,2013,11(3):346-359.
[15] Kang X, Li Y, Qu Z, et al. Enhancing source camera identification performance with a camera reference phase sensor pattern noise[J]. Information Forensics and Security,
2012,7(2):393-402.