An efficient news video browsing system for wireless network application

In this paper, an efficient news video browsing system for wireless network application is presented. To provide a better news video categorization, the news scripts of all the news stories are utilized to perform semantic analysis first. Theses stories are then clustered as a dynamic hierarchical tree using the proposed term-based entropy. To reduce the network loading in wireless environments, this study also presents a key frame extraction method for only displaying the news video abstract. The superiority of our system has been demonstrated through the three days news video obtained from EraNews in Taiwan.