Font Size: a A A

The Research On The Extraction And Retrieval Of Keyword In Network Video Subtitles

Posted on:2018-12-20Degree:MasterType:Thesis
Country:ChinaCandidate:W YuanFull Text:PDF
GTID:2348330515983296Subject:Software engineering
Abstract/Summary:PDF Full Text Request
With the development of science and technology,the computer network becomes an important factor in people's lives.Video,images,animation and other multimedia information have become the dominant design in internet.Subtitles are an important part of the video and are often used to help video viewers understand video content.Subtitles contain rich video information,which is the important basis of video analysis and video retrieval.We can quickly retrieve video content based on the keywords of the video subtitles that we extracted.In the subtitles extraction section,this paper proposes a subtitles quickly locating method based on FAST corner detection algorithm.The method uses the improved FAST corner detection algorithm for corner detection.In the corner image,the horizontal integral projection is used to locate the subtitle line information,the partial morphological processing and the vertical integral projection method are used to locate the subtitle column information,and the subtitle candidate position is obtained.Finally,the subtitle position is obtained by using the heuristic rule.In the subtitles keyword extraction section,this paper proposes a subtitles keyword extraction method based on the improved TF-IDF algorithms.This method proposes to use the part of speech coefficient,the title coefficient and the TF-IDF weight to combine to calculate the keyword eigenvalue.This method improves the accuracy of keyword extraction.In the custom search section,this paper proposes an improved Sunday string matching algorithm for keyword retrieval.In this paper,we use the improved Sunday algorithm(WSunday algorithm)to match the custom keywords and subtitle strings.WSunday algorithm if does not match the current string characters in pattern string of characters,but not bad character,look ahead the characters in the pattern in the location of the string,and compare to the right or left its character,based on the mobile mode window,increase the modal window mobile distance,to lower the number of matching,improve the matching efficiency.In this paper,we use the above method to develop the network video subtitle keyword extraction and retrieval system.The system includes a subtitle extraction module,a subtitle keyword extraction module,a subtitle keyword storage module,and a subtitle keyword retrieval module.The experimental results show that the extraction of keywords in the network video subtitles is good,and the retrieval of video with subtitle keywords is very high.
Keywords/Search Tags:subtitles location, FAST algorithm, keyword extraction, keyword retrieval
PDF Full Text Request
Related items