摘要
目前根据影片结构及MPEG影片压缩技术所发展出来的分镜侦测技术目前已有相当多种可供利用,举例来说在未压缩域有pixel wise comparison,likelihood ratio及histogram comparison; 而在压缩域则有MPEG离散余弦系数及直流值画面等方法。利用分镜侦测技术所发展出来的数字影片内容分析检索技术可应用在许多的领域,可节省在网络及一般影片出版流程上人力所需耗时处理的动作。为此针对几项主要的分镜侦测方法的原理及其优缺点进行说明,并以网络影片数据库及自动化的内容查询系统两项例子来说明其在网络多媒体出版上的应用。
There are many video shot change detection techniques that are based on the feature extraction from the movie structure in raw-data domain and frequency domains such as “pixel-wise comparison', “histogram comparison', “likelihood ratio' , “DCT coefficients” and “DC image'. The automatic video indexing and summarization technique is exceedingly helpful for accessing the video segments in video archive. Among the related techniques, the shot change detection is the most primary technique to achieve the above-mentioned purposes. It described the various video change detection techniques in raw-data and frequency domains, and analyzed the advantages or disadvantages of the mentioned techniques, and then introduced the joint-model video shot change detection approach. Lastly, the multimedia publishing system is presented which utilizes our developed approach to construct the index and summarization of the video.
出处
《仪器仪表学报》
EI
CAS
CSCD
北大核心
2004年第z2期109-112,共4页
Chinese Journal of Scientific Instrument
关键词
影片分镜技术
网络多媒体出版
影片检索
影片摘要
Shot change detection Multimedia publishing Video indexing Video summarization