A Survey on Content based Video Retrieval Using Speech and Text information

Creating video recordings of events such as lectures or meetings is increasingly less expensive and easy. Thus the Video data is increasing in a great deal on World Wide Web (www) and so thus the need of more efficient and correctly functioning method of video indexing, grouping and video retrieval in WWW or Large video archives is necessary. This paper presents a speech and text based video retrieval and Video search system using Optimal Character Recognition (OCR) and Automated Speech Recognition (ASR). First, we convert the video into key-frames and extract the Audio and Text using OCR and ASR. Following step is to produce a summary presenting key points of the video, by making use of text and audio extracted from the Video. This summary will then be used for grouping and Indexing of videos. This in turn will improve the users aptitude to quickly review this material. This will make user go through only information that they needed. However, the text in the video may vary in dimension, orientation, style, background, contrast and variations in rhythm, volume of and noise in speech and the differentiating between the key- speeches and unnecessary other sounds used during the recording as well, makes data extraction extremely challenging.