Research on Fast Video Copy Detection for Copyright Protection Using Adjacent Pixel Intensity Difference Quantization

为了防止未经授权地使用互联网上的视频等数字内容，近年来迫切需要快速、准确地检测大型视频数据库中非法视频副本的技术。传统的算法使用帧图像的归一化灰度特征，因此其结果可能对帧内亮度和运动的微小变化敏感。在这项研究中，我们提出了一个快速和强大的搜索算法的复制检测使用一个新的功能的基础上相邻像素强度差量化（APIDQ）算法。由于这种特性与主动搜索算法兼容，因此将APIDQ与主动搜索相结合也可以实现更快的搜索速度。实验结果表明，该算法比传统的快速视频搜索算法能更准确地检测出相似拷贝视频片段，具有更好的鲁棒性。

In order to prevent unauthorized use of digital contents such as video on the Internet, fast and accurate techniques which can detect illegal video copies from large video databases are strongly required in recent years. In conventional algorithm, normalized intensity features of the frame image are used, so the results may be sensitive to small change of luminance and motion in the frame. In this research, we present a fast and robust search algorithm for copy detection using a new feature based on the Adjacent Pixel Intensity Difference Quantization (APIDQ)algorithm. Because such a feature is compatible with active search algorithm, fast search speed can also be achieved by combining APIDQ and active search. Experimental results show the proposed algorithm can detect the similar copy video clip more accurately and robust than conventional fast video search algorithm.