Research on Automatic Video indexing System Using Adjacent PixelIntensity Difference Quantization

近年来，视频索引已经成为一个活跃的研究领域，因为视频内容在网络上变得司空见惯，视频数据库的大小由于互联网连接和磁盘存储技术的快速发展而迅速增加。在本研究中，我们试图开发一个自动视频索引系统的大型视频数据库。该算法采用相邻像素灰度差量化（APIDQ）直方图作为特征向量。实验结果表明，与传统的快速视频搜索算法相比，该算法能更准确地检测出相似视频片段，且对高斯噪声具有较强的鲁棒性。

Video indexing has become an active area of research in recent years because video content becomes commonplace on the web and the size of video database quickly increases due to rapid developments of internet connection and disk storage technology. In this research, we sought to develop an automatic video indexing system for large video database. The adjacent pixel intensity difference quantization (APIDQ) histogram is used as the feature vector in this algorithm. Experimental results show the proposed algorithm can detect the similar video clip more accurately and robust against Gaussian noise than conventional fast video search algorithm.