A Real-Time Method to Select and Prioritize Frames of a News Program

This paper presents a system that willautomatically recognize the video frames during thebroadcast a typical news program. Such frames representusually images of anchorpersons, guest speakers,commercials and key news images. The system also workson prioritizing the news images per news story, in anattempt to generate an attractive and easy to read pictorialtranscript. The system focuses on selecting the proper keyframesfrom every news clip avoiding selecting imagesfrom commercials. It also works on recognizing theanchorperson image, and distinguishing it from imagesthat belong to guest speakers. The outcome of this systemis a pictorial transcript that is friendly and easy to read,index, and understand, especially when marking newsimages by their importance within each news story.