A Review on: Video Searching Using Speech & Video Text Information

ABSTRACT: Since Last decade e-lecturing has become more and more popular. There is a continuous increment in the number of lecture videos data on the Internet and there is no doubt that in near future it will be booming field of e-lectures. Hence there is a need for highly flexible and best results-yielding algorithms or methods to search appropriate contents. Thus, indexing is an approach which is mainly emphasized in this paper for automated video indexing and video search in large lecture video archives. Primarily, Videos are divided into some segments and then key-frames are identified among these segments for further protocols of video content data extraction and mining. Immediately after that, steps like extraction of textual metadata by applying Optical Character Recognition (OCR) technology on key-frames of videos and Automatic Speech Recognition (ASR) on audio lecture tracks are followed. The OCR and ASR transcripts are the methods from which textual data is extracted and stored into digitized format and can be used as keywords related to videos, matching these keywords with the available database of videos on the Internet and thus will provide relevant Videos and contents. The evaluation of results for performance in index based searching method is remarkable as compared to other searching methods. Index Terms: Lecture videos, automatic video indexing, content-based video search, lecture video archives I