摘要：Digital video is getting increasingly common, and many broadcasters consider switching to digital formats. Video data requires appropriate indices to be available for effective use. Manual indexing methods are ineffective in high volume databases. This article is an introduction to the basics of automatic parsing of news videos, the fi rst step of content based indexing. The article goes through shot detection methods and presents episode extraction by modelling, the two fundamental phases of recovering news video structure. Also, a news broadcast from Finnish Broadcast Company (YLE) is considered in the light of these content based segmentation methods.
