数字图像区域复制粘贴特征点盲检测方法仿真  

Digital Image Area Copy and Paste Feature Point Blind Detection Method Simulation

作　　者：魏佳[1] 

机构地区：[1]陕西理工大学数学与计算机科学学院,陕西汉中723001

出　　处：《计算机仿真》2017年第12期260-263,共4页Computer Simulation

基　　金：陕西省教育厅科研专项(14JK1144);秦巴山区生物资源综合开发协同创新中心自然科学项目(QBXT-Z(P)-15-20)

摘　　要：为了准确检测数字图像复制粘贴的篡改区域,抵御后续各种后处理操作,需要对复制粘贴区域进行盲检测,采用当前盲检测方法对压缩、加噪等处理下的数字图像难以提取特征点,存在计算量大、检测精度不高的问题,提出一种基于特征提取的数字图像区域复制粘贴特征点盲检测方法。上述方法首先提取数字图像SIFT关键点,再对各个图像关键点生成中心对称局部二进制特征描述子,对生成的数字图像特征描述符进行重新排序,采用g2NN算法对数字图像特征点进行匹配,计算出复制粘贴的源数字图像区域和目标区域之间的几何形变关系,由此完成复制粘贴区域精确定位检测。实验结果表明,所提方法对压缩、加噪等操作具有较强的鲁棒性,检测精度较高。This paper puts forward a blind detection method for copying and pasting feature point in digital image region based on feature extraction. SIFTs key point of digital image is extracted and local binary system feature descriptor with central symmetry is generated, then the generated feature descriptor of the digital image is reordered and g2NN algorithm is used to match feature point of the digital image. The geometric deformation relationship between copying and pasting origin digital image region and object region is worked out. Thus, the accurate location detection of copying and pasting region is completed. Simulation results show that the method has strong robustness in aspect of operation of compressing and adding noise and has high detection precision.

关 键 词：数字图像区域 复制粘贴特征点 盲检测方法 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]