Electronic Media

The Research & Development of Technology Division of the National Digital Archives Program has developed advanced technologies for managing digital video archives. With these technologies, we can build indexing systems for the fast retrieval of digital video content, and also add value to applications of the content.

The architecture of our digital video archive system is comprised of the following processes: 1) collating films, 2) capturing digital video content, 3) processing and analyzing video content, 4) building databases, and 5) adding value to digital content applications. We have established a complete digital video archive system by integrating and developing the technologies of video format transformation, video shot detection, video abstract extraction, key frame extraction, metadata searching, full text searching, voice searching, streaming video format, and a nonlinear online editing tool. The workflow of the system is shown in Fig. 1, and the user interface for the web page of video searching is shown in Fig. 2.




Fig.1 Workflow for the digital video archive system




Fig.2 User interface for the video searching web page

The major technologies used in our digital video archive system are:

1.	Video format transformation: This is used to transform video file formats ﹙MPEG1/2, RM, WMV﹚and digital content features ﹙frame size, bit rate﹚into different formats. For example, we can transfer a digital video in MPEG2 format to one in WMV format with a 720*480 frame size.
2.	Shot detection and abstract extraction: This is used to detect time points where shots change dramatically. The detection results are then used to extract video abstracts and static images, which allows users to quickly preview the video content.
3.	Voice searching: This recognizes voice content from a video file automatically so that voice searching is feasible, even if content providers do not provide metadata of the video.
4.	Non-linear online editing: This allows users to mark and save video segments while previewing video content online, and view the segments again later. The technology could also be applied to managing online trading of video segments; for example, a user could buy the segment containing the 5th to the 10th minute of a video. The user interface for the non-linear online editing tool is shown in Fig. 3.




Fig.3 User interface of the nonlinear online editing tool

Our research has been applied to the Digital Museum of Taiwan's Social and Humanities Video Archive system, which contains more than 3,000 hours of film about Taiwan’s society and humanities. We have also collaborated with Wordpedia Co. Ltd in experiments on value-added applications, and deploying business operation models on video content. The resulting technologies have been used to build an audio/video database for the Government Information Office. Finally, some of the technologies we have developed for digital video archives have been licensed to HyWeb Co. Ltd.