Abstract
In this paper, we propose a new method for searching and browsing news videos, based on multi-modal approach. In the proposed scheme, we use closed caption (CC) data to index the contents of TV news articles effectively. To achieve time alignment between the CC texts and video data, which is necessary for multi-modal search and visualization, supervised speech recognition technique is employed. In our implementations, we provide two different mechanisms for news video browsing. One is to use a textual query based search engine, and the other is to use topic based browser which acts as an assistant tool for finding the desired news articles. Compared to other systems mainly dependent on visual features, the proposed scheme could retrieve more semantically relevant articles quite well.
Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
