基于背景建模和屬性學(xué)習(xí)的視頻摘要研究
發(fā)布時(shí)間：2018-10-16 16:33
【摘要】：隨著高清攝像設(shè)備的普及和物聯(lián)網(wǎng)的興起以及平安城市和智慧城市的提出,監(jiān)控?cái)z像頭被廣泛地部署在城市的每一個(gè)角落。監(jiān)控設(shè)備可以在打擊違法犯罪,維護(hù)社會(huì)長(zhǎng)治久安上發(fā)揮重要的作用。然而海量的視頻數(shù)據(jù)也在視頻的存儲(chǔ)歸檔和查閱檢索上給人們帶來巨大的考驗(yàn)。傳統(tǒng)的直接存儲(chǔ)和人工檢索方式已經(jīng)無法應(yīng)對(duì)大規(guī)模視頻的處理需求。如何解決海量視頻的存儲(chǔ)和檢索的難題已經(jīng)成為國內(nèi)外學(xué)者研究的熱點(diǎn)。因此本文針對(duì)這兩個(gè)難題展開了相關(guān)研究。在查閱了大量國內(nèi)外文獻(xiàn)和資料之后,對(duì)視頻存儲(chǔ)和檢索領(lǐng)域有了一定的了解,深入分析了課題的研究現(xiàn)狀。闡述了當(dāng)前研究工作的主要難點(diǎn)在于如何將監(jiān)控視頻中前景對(duì)象準(zhǔn)確且無遺漏地檢測(cè)出來;在檢測(cè)出前景后如何對(duì)其進(jìn)行多概念檢測(cè);在對(duì)多概念對(duì)象進(jìn)行分類和描述時(shí)如何跨越語義鴻溝等。在此基礎(chǔ)上本文提出了基于背景檢測(cè)和屬性學(xué)習(xí)的視頻摘要方法。利用改進(jìn)后的ViBe對(duì)視頻序列進(jìn)行背景建模,去除不包含前景對(duì)象的視頻幀,將其余幀保留下來生成濃縮后的視頻,以達(dá)到減少視頻文件對(duì)存儲(chǔ)造成的壓力的目的;在獲取到前景對(duì)象后建立屬性分類器,利用屬性學(xué)習(xí)對(duì)前景對(duì)象進(jìn)行概念檢測(cè),檢測(cè)出相應(yīng)概念后利用屬性標(biāo)簽來描述該前景對(duì)象,由此在濃縮的視頻基礎(chǔ)上生成視頻摘要。本文研究的主要內(nèi)容如下:(1)提出了基于改進(jìn)ViBe的視頻背景建模與濃縮。在對(duì)視頻背景建模算法進(jìn)行研究對(duì)比后,選擇較其他主流方法速度快、占用內(nèi)存少的ViBe算法。針對(duì)原ViBe算法在實(shí)際監(jiān)控場(chǎng)景下仍會(huì)存在噪點(diǎn)和閃爍點(diǎn)以及在初始化過程中會(huì)引入鬼影的問題,對(duì)ViBe算法進(jìn)行改進(jìn),分別提出了基于計(jì)數(shù)點(diǎn)閾值的閃爍點(diǎn)去除方法,基于形態(tài)學(xué)的噪點(diǎn)消除方法,和面向鬼影區(qū)域檢測(cè)和抑制的改進(jìn)算法。在實(shí)現(xiàn)并實(shí)驗(yàn)驗(yàn)證了對(duì)ViBe的改進(jìn)后,將其應(yīng)用于前景提取與視頻濃縮中去。首先對(duì)視頻進(jìn)行背景建模,獲取前景對(duì)象。而后將不包含前景對(duì)象的無用幀略去,以達(dá)到去除時(shí)間維度上的冗余信息的目的,對(duì)視頻進(jìn)行濃縮。(2)提出了基于多核屬性學(xué)習(xí)的前景多概念檢測(cè)與摘要。首先將多核學(xué)習(xí)引入直接屬性預(yù)測(cè)模型框架中,給出了對(duì)核函數(shù)的權(quán)重向量進(jìn)行優(yōu)化求解方法;進(jìn)一步地,將提出的模型運(yùn)用視頻對(duì)象分類中;繼而利用模型的多概念分類能力和屬性描述能力,對(duì)監(jiān)控視頻前景多概念進(jìn)行檢測(cè),并給檢測(cè)出的對(duì)象加上屬性標(biāo)簽,生成視頻摘要;最后,設(shè)計(jì)對(duì)比實(shí)驗(yàn)對(duì)提出方法的有效性進(jìn)行驗(yàn)證。(3)在前面兩個(gè)研究點(diǎn)的基礎(chǔ)上,運(yùn)用軟件工程中面向?qū)ο蟮乃悸反罱ɑ诒尘敖：蛯傩詫W(xué)習(xí)的視頻摘要原型系統(tǒng)。系統(tǒng)包含視頻濃縮模塊、屬性預(yù)測(cè)模型訓(xùn)練模塊、視頻摘要模塊。運(yùn)行效果良好,達(dá)到了本研究的預(yù)期目標(biāo)。
[Abstract]:With the popularization of high-definition camera equipment and the rise of Internet of things and the introduction of Ping'an City and Smart City, surveillance cameras are widely deployed in every corner of the city. Monitoring equipment can play an important role in cracking down on crime and maintaining social stability. However, the huge amount of video data also brings people a great test in the storage, archiving and retrieval of video. Traditional methods of direct storage and manual retrieval can no longer cope with the need of large-scale video processing. How to solve the problem of mass video storage and retrieval has become a hot topic for scholars at home and abroad. Therefore, this paper has carried out the related research in view of these two difficult problems. After consulting a large number of domestic and foreign literature and materials, we have a certain understanding of video storage and retrieval field, in-depth analysis of the research status of the subject. The main difficulties of the current research work are how to detect the foreground objects accurately and without omission, how to detect the foreground objects accurately and how to detect them with multiple concepts after detecting the foreground. How to cross the semantic gap when classifying and describing multi-concept objects. On this basis, this paper proposes a video summarization method based on background detection and attribute learning. The improved ViBe is used to model the background of the video sequence, remove the video frames without foreground objects, and save the remaining frames to generate the condensed video, so as to reduce the pressure caused by the video files on the storage. After obtaining the foreground object, the attribute classifier is established, and the concept of foreground object is detected by using attribute learning, and then the foreground object is described by attribute label, and the video summary is generated on the basis of condensed video. The main contents of this paper are as follows: (1) the video background modeling and concentration based on improved ViBe is proposed. After studying and comparing the video background modeling algorithm, the ViBe algorithm, which is faster than other mainstream methods and occupies less memory, is selected. In view of the problem that the original ViBe algorithm still has noise and flicker points in the actual monitoring scene and the ghosts will be introduced in the initialization process, the ViBe algorithm is improved, and the method of removing the flashing points based on the count point threshold is proposed respectively. Morphology based noise cancellation method, and an improved algorithm for ghost region detection and suppression. After the implementation and experimental verification of the improved ViBe, it is applied to foreground extraction and video concentration. Firstly, the background of the video is modeled and the foreground object is obtained. Then the useless frame without foreground object is omitted to remove redundant information in time dimension and the video is condensed. (2) Multi-concept detection and summary of foreground based on multi-core attribute learning is proposed. Firstly, multi-kernel learning is introduced into the framework of direct attribute prediction model, and the optimization method of weight vector of kernel function is given. Furthermore, the proposed model is applied to video object classification. Then, the multi-concept classification ability and attribute description ability of the model are used to detect the multi-concept of the surveillance video foreground, and the detected objects are tagged with attributes to generate the video summary. A comparative experiment is designed to verify the effectiveness of the proposed method. (3) on the basis of the above two research points, a video abstract prototype system based on background modeling and attribute learning is built by using the object-oriented approach in software engineering. The system includes video enrichment module, attribute prediction model training module and video summary module. The operation effect is good and the expected goal of this study has been achieved.
【學(xué)位授予單位】：江蘇大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2017
【分類號(hào)】：TP391.41
本文編號(hào)：2274982
[Abstract]:With the popularization of high-definition camera equipment and the rise of Internet of things and the introduction of Ping'an City and Smart City, surveillance cameras are widely deployed in every corner of the city. Monitoring equipment can play an important role in cracking down on crime and maintaining social stability. However, the huge amount of video data also brings people a great test in the storage, archiving and retrieval of video. Traditional methods of direct storage and manual retrieval can no longer cope with the need of large-scale video processing. How to solve the problem of mass video storage and retrieval has become a hot topic for scholars at home and abroad. Therefore, this paper has carried out the related research in view of these two difficult problems. After consulting a large number of domestic and foreign literature and materials, we have a certain understanding of video storage and retrieval field, in-depth analysis of the research status of the subject. The main difficulties of the current research work are how to detect the foreground objects accurately and without omission, how to detect the foreground objects accurately and how to detect them with multiple concepts after detecting the foreground. How to cross the semantic gap when classifying and describing multi-concept objects. On this basis, this paper proposes a video summarization method based on background detection and attribute learning. The improved ViBe is used to model the background of the video sequence, remove the video frames without foreground objects, and save the remaining frames to generate the condensed video, so as to reduce the pressure caused by the video files on the storage. After obtaining the foreground object, the attribute classifier is established, and the concept of foreground object is detected by using attribute learning, and then the foreground object is described by attribute label, and the video summary is generated on the basis of condensed video. The main contents of this paper are as follows: (1) the video background modeling and concentration based on improved ViBe is proposed. After studying and comparing the video background modeling algorithm, the ViBe algorithm, which is faster than other mainstream methods and occupies less memory, is selected. In view of the problem that the original ViBe algorithm still has noise and flicker points in the actual monitoring scene and the ghosts will be introduced in the initialization process, the ViBe algorithm is improved, and the method of removing the flashing points based on the count point threshold is proposed respectively. Morphology based noise cancellation method, and an improved algorithm for ghost region detection and suppression. After the implementation and experimental verification of the improved ViBe, it is applied to foreground extraction and video concentration. Firstly, the background of the video is modeled and the foreground object is obtained. Then the useless frame without foreground object is omitted to remove redundant information in time dimension and the video is condensed. (2) Multi-concept detection and summary of foreground based on multi-core attribute learning is proposed. Firstly, multi-kernel learning is introduced into the framework of direct attribute prediction model, and the optimization method of weight vector of kernel function is given. Furthermore, the proposed model is applied to video object classification. Then, the multi-concept classification ability and attribute description ability of the model are used to detect the multi-concept of the surveillance video foreground, and the detected objects are tagged with attributes to generate the video summary. A comparative experiment is designed to verify the effectiveness of the proposed method. (3) on the basis of the above two research points, a video abstract prototype system based on background modeling and attribute learning is built by using the object-oriented approach in software engineering. The system includes video enrichment module, attribute prediction model training module and video summary module. The operation effect is good and the expected goal of this study has been achieved.
【學(xué)位授予單位】：江蘇大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2017
【分類號(hào)】：TP391.41
【參考文獻(xiàn)】
相關(guān)期刊論文 前10條
1 魏廣巨;;視頻監(jiān)控行業(yè)發(fā)展觀察[J];中國安防;2016年11期
2 崔桐;徐欣;;一種基于語義分析的大數(shù)據(jù)視頻標(biāo)注方法[J];南京航空航天大學(xué)學(xué)報(bào);2016年05期
3 蘭紅;周偉;齊彥麗;;動(dòng)態(tài)背景下的稀疏光流目標(biāo)提取與跟蹤[J];中國圖象圖形學(xué)報(bào);2016年06期
4 鄧仕超;黃寅;;二值圖像膨脹腐蝕的快速算法[J];計(jì)算機(jī)工程與應(yīng)用;2017年05期
5 王輝;宋建新;;一種基于閾值的自適應(yīng)Vibe目標(biāo)檢測(cè)算法[J];計(jì)算機(jī)科學(xué);2015年S1期
6 何曦;瞿建榮;盧曉燕;劉培楨;王嬌穎;;基于幀間差分和水平集的運(yùn)動(dòng)目標(biāo)探測(cè)跟蹤方法[J];探測(cè)與控制學(xué)報(bào);2015年01期
7 王娟;蔣興浩;孫錟鋒;;視頻摘要技術(shù)綜述[J];中國圖象圖形學(xué)報(bào);2014年12期
8 楊勇;孫明偉;金裕成;;一種改進(jìn)視覺背景提取(ViBe)算法的車輛檢測(cè)方法[J];重慶郵電大學(xué)學(xué)報(bào)(自然科學(xué)版);2014年03期
9 胡小冉;孫涵;;一種新的基于ViBe的運(yùn)動(dòng)目標(biāo)檢測(cè)方法[J];計(jì)算機(jī)科學(xué);2014年02期
10 王文豪;周泓;嚴(yán)云洋;;一種基于連通區(qū)域的輪廓提取方法[J];計(jì)算機(jī)工程與科學(xué);2011年06期
相關(guān)碩士學(xué)位論文 前2條
1 詹智財(cái);基于卷積神經(jīng)網(wǎng)絡(luò)的視頻語義概念分析[D];江蘇大學(xué);2016年
2 馮嘉;SIFT算法的研究和改進(jìn)[D];吉林大學(xué);2010年
，本文編號(hào)：2274982
本文鏈接：http://www.lk138.cn/shoufeilunwen/xixikjs/2274982.html
最近更新
教材專著
