基于内容的视频拷贝检测是多媒体领域的一个研究热点。由于拷贝变换的多样性和综合性,单一特征难以获得很好的检测效果。提出一种多特征综合的方法来提高视频拷贝检测的效果。除了使用传统的局部和全局视觉特征外,还使用非正交二值子空间(NBS)方法来表示视频内容,并在其基础上使用归一化互相关(NCC)来提高拷贝视频内容相似度计算的效果。在此基础上,还采用多种措施对拷贝视频的判定结果进行精化。实验结果表明,该套方案对多种拷贝变换具有很强的鲁棒性,并且能够得到很好的检测精度。

Nowadays, content based video copy detection has become a widely studied issue. Due to the uncertainty and diversity of video copy transformation, it is difficult to achieve great performance based on single visual features in video copy detection. In this paper, we propose a new method, which uses a multiple visual feature synthesizing method to solve the problem. Besides traditional local and global visual features, we additionally employ nonorthogonal binary subspace(NBS) as special visual feature to represent the video content. On this basis, the normalized cross correlation(NCC) is used to improve the performance of similarity matching of the video content. We also used other measures to improve the detection accuracy. The experiment results show that our system is robust to various video transformations, and achieves better detection accuracy.