中文摘要：

该文提出一种基于多特征集成的鲁棒图像拷贝检测算法。该算法结合局部与全局特征信息，综合时域与频域信号处理，采用级联式检测过滤框架。为了在大规模图像库下提升拷贝检测速度，利用K—D树建立特征索引，便于图像之间的局部点匹配。最后，利用RANSAC算法，剔除错误的匹配对。实验结果证明，该算法对拉伸、平移、旋转、加噪、水彩、滤波和JPEG压缩等攻击具有鲁棒性，并具有较高检测效率。

英文摘要：

This paper proposes an algorithm based on the integrated of multi-feature robust image copy detec- tion. This algorithm combines local feature with global feature information, combines time domain signal pro- cessing with frequency domain signal processing, and uses the cascade filtering framework. To improve the speed of image copy detection under massive image lib, the algorithm uses K-D tree to establish feature index. Last, the algorithm uses RANSAC algorithm to eliminate the false matching pair. Experimental results show that the algorithm is robust under the attack of stretching, translation, rotation, noise, watercolor, filtering and JPEG compression with high detection efficiency.