Automated Extraction of Lecture Outlines from Lecture Videos - A Hybrid Solution for Lecture Video Indexing

Multimedia-based tele-teaching and lecture video portals have become more and more popular in the last few years. The amount of multimedia data available on the WWW (World Wide Web) is rapidly growing. Thus, finding lecture video data on the web or within a lecture video portal has become a significant and challenging task. In this paper, we present an approach for lecture video indexing based on automated video segmentation and extracted lecture outlines. First, we developed a novel video segmenter intended to extract the unique slide frames from the lecture video. Then we adopted video OCR (Optical Character Recognition) technology to recognize texts in video. Finally, we developed a novel method for extracting of lecture outlines from OCRtranscripts. Both video segments and extracted lecture outlines are further utilized for the video indexing. The accuracy of the proposed approach is proven by evaluation.