Researches On Quick Browsing Approaches To Academic Lecture Videos In University Digital Library

At present, the lecture video has become an important way to acquire knowledge. However, with more and more lecture videos, it is a problem, which we should solve urgently, for the readers to browse the video which they are interested in quickly and locate to the specific clips. So the quick browse technology emerged, and has become one of hotspots in the area of video processing.Key frames-based video abstract and video caption extraction are two important technologies to achieve quick video browsing. Based on the previous researches, this thesis presents a content based scheme to browse lecture videos quickly, and a key frames and key words based scheme to locate clips in the video of academic lectures. The main contributions are as follows:1. A key frame extraction algorithm based on lecture video is presented. According to that the shots in the lecture videos change abruptly, the key frame extraction depends on the content of the shots in this algorithm.2. A key frame extraction algorithm based on caption detection is presented. Considering that the caption can describe the video content exactly, the frame difference method is used to delete the frames without caption in this algorithm.3. A key frame extraction algorithm based on shots and caption comparison is presented. Because the captions are at the bottom of the video frames, the frame difference method is only needed to be used to compare the bottom of the video frames in order to delete the video frames with same caption.4. A system for quick browsing the lecture video based on key frames and caption location is proposed. In the system, the interested shots can not only be localized according to selection of key frames, but also be localized based on the key words included in the caption.