本文關(guān)鍵詞：基于Hadoop架構(gòu)的分布式視頻關(guān)鍵幀提取方法研究　出處：《合肥工業(yè)大學(xué)》2016年碩士論文　論文類型：學(xué)位論文

  更多相關(guān)文章： 關(guān)鍵幀 視頻大數(shù)據(jù) 云計(jì)算 分布式計(jì)算

【摘要】：隨著視頻監(jiān)控技術(shù)在各行業(yè)領(lǐng)域內(nèi)被廣泛運(yùn)用,導(dǎo)致了監(jiān)控視頻數(shù)據(jù)量呈指數(shù)增長,人們早已經(jīng)厭煩了以往根據(jù)時(shí)間信息手動(dòng)拖拽進(jìn)度條的瀏覽方式,不僅消耗了查詢者大量的時(shí)間和精力,而且容易遺漏關(guān)鍵目標(biāo)信息。關(guān)鍵幀提取技術(shù)的出現(xiàn)使得這種情況得到了極大的改善,關(guān)鍵幀可以不受時(shí)間、音視頻同步等問題的影響,并可以提供多種方式進(jìn)行瀏覽和導(dǎo)航使用。但是在實(shí)際應(yīng)用關(guān)鍵幀提取技術(shù)的過程中,面臨的最大困難就是關(guān)鍵幀提取速度太慢：一方面是算法的高復(fù)雜性；另一方面是當(dāng)前使用的是單機(jī)模式提取算法。本文在對(duì)關(guān)鍵幀提取算法充分調(diào)研之后,提出一種新的關(guān)鍵幀提取方法,并在將算法移植到Hadoop云平臺(tái)的過程中,解決了幀完整性等核心問題,成功將算法改造成了分布式提取方式,極大地提高了關(guān)鍵幀的提取速度。首先,本文指出了關(guān)鍵幀技術(shù)對(duì)視頻檢索等實(shí)際工程應(yīng)用的重要性,并介紹了近年來發(fā)展如火如荼的云計(jì)算技術(shù)和其在多媒體領(lǐng)域的應(yīng)用現(xiàn)狀,同時(shí)對(duì)Hadoop的核心技術(shù)和理論進(jìn)行了充分的學(xué)習(xí)和調(diào)研。其次,本文分析了當(dāng)前關(guān)鍵幀提取方法存在的問題,提出一種新的自適應(yīng)的視頻關(guān)鍵幀提取方法,該方法能夠自適應(yīng)確定關(guān)鍵幀數(shù)目、計(jì)算量小并且對(duì)內(nèi)容漸變的視頻的處理效果更佳。再次,本文分析了在Hadoop云平臺(tái)上實(shí)現(xiàn)視頻關(guān)鍵幀的分布式提取云應(yīng)用所面臨的難題,例如幀的完整性、處理邏輯的MapReduce化等問題,并給出了有效的解決方案。最后,本文在上文的理論支持和對(duì)云應(yīng)用開發(fā)的相關(guān)技術(shù)充分調(diào)研的基礎(chǔ)之上,搭建了Hadoop分布式集群環(huán)境,并且實(shí)現(xiàn)了基于Hadoop云平臺(tái)的分布式視頻關(guān)鍵幀提取系統(tǒng).實(shí)驗(yàn)結(jié)果表明,本文算法能夠很大程度的提高關(guān)鍵幀的提取速度,與單機(jī)提取模式相比,更加適合處理視頻大數(shù)據(jù)。
[Abstract]:With the video surveillance technology has been widely used in various industries, to monitor the amount of video data is exponential growth, it has already tired of the previous time according to the information manually drag the progress bar to browse, query not only consumes a lot of time and energy, but also easy to miss the target information. Key technology of key frame extraction the situation has been greatly improved, the key frames can be not affected by time, influence of synchronization of audio and video, and can provide a variety of ways for browsing and navigation. But in the actual application process of extraction of key frames, the biggest difficulty is the key frame extraction speed is too slow: on the one hand is the high complexity algorithm; on the other hand is the current use of method is stand-alone mode. Based on the full investigation of key frame extraction algorithm, put forward a new turn Key frame extraction method, and the algorithm is transplanted to the Hadoop cloud platform, solve the frame integrity of the core issues, success will be transformed into a distributed algorithm extraction method, greatly improving the extraction rate of key frame. Firstly, this paper points out the importance of practical engineering technology of key frame of video retrieval applications in recent years, and introduces the development like a raging fire of cloud computing technology and its application in the field of multimedia, the core technology and theory of Hadoop are full of learning and research. Secondly, this paper analyzes the current problems of key frame extraction method, video key frame extraction method is proposed and a new adaptive. This method can adaptively determine the number of key frames, better treatment effect and small amount of calculation of the content gradient of the video. Thirdly, this paper analyzes on the Hadoop cloud platform to realize the video off Distributed key frame extraction problem faced by cloud applications, such as frame integrity, MapReduce issues such as processing logic, and gives effective solutions. Finally, based on the above theoretical support and related technology of cloud application development of full investigation, set up a Hadoop distributed cluster environment, and the implementation of distributed video key frame extraction system based on Hadoop cloud platform. The experimental results show that this algorithm can improve the speed of extracting key frames greatly, compared with the single extraction mode, more suitable for processing video data.

【學(xué)位授予單位】：合肥工業(yè)大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2016
【分類號(hào)】：TN948.6;TP393.09

【相似文獻(xiàn)】

相關(guān)期刊論文 前10條

1 于俊清,周洞汝,劉軍,蔡波;基于文字和圖像信息提取視頻關(guān)鍵幀[J];計(jì)算機(jī)工程與應(yīng)用;2002年09期

2 張新春;基于運(yùn)動(dòng)活力的視頻分鏡中關(guān)鍵幀的提取[J];電子與電腦;2005年03期

3 陳丹雯;張俊;韓兵;吳玲達(dá);;基于改進(jìn)詞袋模型的相似關(guān)鍵幀匹配方法[J];計(jì)算機(jī)工程與設(shè)計(jì);2011年08期

4 劉善磊;趙銀娣;王光輝;李英成;薛艷麗;李建軍;;一種關(guān)鍵幀的自動(dòng)提取方法[J];測繪科學(xué);2012年05期

5 王毅霞;崔大力;;探究工程監(jiān)理系統(tǒng)中關(guān)鍵幀的提取技術(shù)[J];甘肅冶金;2013年06期

6 王穎志;淺析關(guān)鍵幀動(dòng)畫曲線的應(yīng)用[J];電視字幕(特技與動(dòng)畫);2002年07期

7 楊潤珍;傅電仁;;運(yùn)動(dòng)圖片檔案的檢索[J];陰山學(xué)刊(自然科學(xué)版);2003年01期

8 田文彬;After Effects常用快捷鍵集錦(三)[J];電視字幕(特技與動(dòng)畫);2005年05期

9 馮德旺;蘭建容;;基于關(guān)鍵幀的視像檢索研究[J];福建工程學(xué)院學(xué)報(bào);2007年04期

10 唐自力;馬彩文;李s，

本文編號(hào)：1385906