作者: B. Erol , J.J. Hull
DOI: 10.1109/ACSSC.2003.1291876
关键词:
摘要: Systems for recording presentations are becoming commonly available. Commercial solutions include authoring tools that let users create online representations by audio, video, and presentation slides while a talk is being given. A typical collection of recordings may contain hundreds even thousands recordings, making it very difficult to retrieve particular find specific points in the presentation. This paper describes retrieval technique utilizes digital camera pictures taken during presentations. An enabling image matching algorithm also described. The text layout captured different multimedia streams documents. Experimental results show our method yields high accuracy.