您的位置：首页 > 论文页面
copy-move盲检测中SIFT与SURF算法比较
发表时间：2017-02-28 浏览量：1790 下载量：416
| 全部作者： | 王心怡，杨正洪，张伟伟，牛少彰 |
| 作者单位： | 中国农业大学理学院；北京邮电大学计算机学院 |
| 摘 要： | 为比较和分析尺度不变特征转换（scale-invariant feature transform，SIFT）和快速鲁棒特征（speeded up robust features，SURF）这两种特征算法对图像copy-move 篡改及后续修饰（模糊、缩放、旋转）的盲检测速度和匹配点效果，利用Photoshop 等图像处理软件对图片进行篡改处理，再分别使用两种算法进行实验研究。结果表明，SURF 算法的检测速度是SIFT 算法的3～7 倍，并且在图像经过模糊篡改后的检测效果也优于SIFT算法，但在图像copy-move 后经过缩放和旋转篡改的检测精度却低于SIFT 算法。因此，应根据实际场合中精度和实时性的需求，选用合适的算法进行检测。 |
| 关 键 词： | 计算机应用；图像取证；copy-move 篡改；盲检测；尺度不变特征转换算法；快速鲁棒特征算法 |
| Title： | Comparison of SIFT and SURF algorithms on copy-move by blind detection |
| Author： | WANG Xinyi, YANG Zhenghong, ZHANG Weiwei, NIU Shaozhang |
| Organization： | School of Science, China Agricultural University; School of Computer, Beijing University of Posts and Telecommunications |
| Abstract： | In order to compare and analyze the blind detection speed and matching points when the image has been tampered with copy-move and subsequent modified (blur, rotate, scale) by the two feature algorithms, scale-invariant feature transform (SIFT) and speeded up robust features (SURF) algorithm. Photoshop and other image processing software are used to tamper image, then the two algorithms are used to carry out experimental research. Results showed that the detection rate of SURT algorithm is 3 to 7 times faster than that of SIFT algorithm, and the detetion effect is also better than the SIFT algorithm after the image is tampered with blur. But the detection accuracy is not as good as the SIFT algorithm when the image is tampered with scaling and rotation after copy-move. Therefore, in accordance with the requirements of accuracy and real-time under practical situations, the appropriate algorithm is selected to detdct. |
| Key words： | computer application; image forensics; copy-move forgery; blind detection; scale-invariant feature transform algorithm; speeded up robust features algorithm |
| 发表期数： | 2017年2月第4期 |
| 引用格式： | 王心怡，杨正洪，张伟伟，等. copy-move盲检测中SIFT与SURF算法比较[J]. 中国科技论文在线精品论文，2017，10（4）：404-411. |
1
评论数 0请您登录
暂无评论
下载全文
多维论文