Feature analysis and comparison in quaternion wavelet domain for forensics

摘要：

　　In this paper, a novel forensics feature is proposed, which is based on quatermion wavelet transform (QWT).Comrpared to digital wavelet transform (DWT) and contourlet wavelet trausform (CWT), QWT can provide more feature inforation to identify photorealisitic (PR) and computer generated (CG) images, including one magnitude and three angles of QWT.In the conditon ofthe same model as Farid”s, the proposed scheme can boost the performance of identification.It is shown in the comparison experiments that the detection accuracy of the proposed scherme improve about 10 percent than Farid”s, about 5 percent than (O)zparlak”s.

关键词： QWT CWT DWT feature comparison Forensics

作者: Jinwei Wang Ting Li Yunqing Shi Guanshuo Xu Jingyu Ye

作者单位: Nanjing University of Information Science and Technology, Jiangsu, 210044;New Jersey Institute of Te Nanjing University of Information Science and Technology, Jiangsu, 210044;Jiangsu Monitoring Center New Jersey Institute of Technology, Newark, NJ, 07103, US

会议类型: 国内会议

会议名称: 第十二届全国信息隐藏暨多媒体信息安全学术大会

会议地点: 武汉

会议语种:英文

页码: 549-559

在线出版日期: 2015-03-28（万方平台首次上网日期，不代表论文的发表时间）