Paper Title
An Improved Text Extraction Algorithm For Intelligent Search In Videos

Abstract
With the extensive advancement in multimedia technologies across the globe, usage of internet archival and storage of multimedia files has become a challenge. Accessing specific segments of videos is the competent problem addressed in this paper. A solution to this problem requires designing an efficient video retrieval tool with indexing, content analysis, and intelligent search inorder to retrieve the needed video portions by the users. The scope of this solution is to develop a web based application for detecting text and captions from videos. Considering the fact that corners exist more in text and captions, corner based approach methodology is used. The moving caption patterns are detected by the combination of text and motion features. Proposed solution is built in such a way so as to capture the text and motions in any direction across the screen. Text recognition is done by using Optical Character Recognition. Miss rate, recall rate and false rate are the evaluation metrices used to evaluate the efficiency of this work. Experimental results have shown that the proposed approach is very efficient in text detection in comparison with the existing real time ones.