【摘要】：隨著數(shù)字媒體技術(shù)、通信技術(shù)及網(wǎng)絡(luò)技術(shù)的飛速發(fā)展,以視頻為代表的數(shù)字媒體信息的數(shù)量急劇膨脹。短視頻是一類內(nèi)容龐雜的視頻數(shù)據(jù),如何在海量短視頻數(shù)據(jù)中尋找到有效信息一直是用戶關(guān)注的問題,由此產(chǎn)生了視頻索引、視頻檢索等相關(guān)應(yīng)用。視頻標(biāo)注就是解決這些應(yīng)用的核心環(huán)節(jié)。目前視頻標(biāo)注已成為數(shù)字媒體應(yīng)用和計(jì)算機(jī)視覺領(lǐng)域中的一個(gè)熱點(diǎn)研究課題。從語義的角度,視頻可以分割成若干種語義單位。不同的語義單位具有不同的語義內(nèi)涵,在每個(gè)語義層次上均可實(shí)現(xiàn)語義標(biāo)注。本文在對視頻結(jié)構(gòu)進(jìn)行深入分析的基礎(chǔ)上,對視頻片段進(jìn)行分割,形成不同的語義單位,并在鏡頭語義層、場景語義層對短視頻進(jìn)行標(biāo)注。本文的研究成果與創(chuàng)新點(diǎn)主要有:(1)結(jié)合視頻幀的全局特征和局部特征,提出了一種新的結(jié)合視頻動(dòng)態(tài)紋理和SIFT特征的鏡頭邊緣檢測方法。該方法首先對相鄰兩幀圖像進(jìn)行均勻分塊,在RGB顏色空間下,計(jì)算幀中每個(gè)圖像塊的平均梯度。由所有圖像塊的平均梯度形成視頻動(dòng)態(tài)紋理,比較相鄰幀圖像的動(dòng)態(tài)紋理,并結(jié)合相鄰幀SIFT特征的匹配情況來判斷鏡頭的變化。該算法對不同類型的視頻數(shù)據(jù)進(jìn)行鏡頭邊緣檢測,均能取得較高的檢測準(zhǔn)確率。(2)提出一種基于鏡頭事件的視頻語義標(biāo)注模型。在分析視頻結(jié)構(gòu)的基礎(chǔ)上,提取鏡頭中的運(yùn)動(dòng)目標(biāo)和鏡頭關(guān)鍵幀的背景顏色特征來表達(dá)一個(gè)鏡頭的事件,進(jìn)一步延伸到場景事件的表達(dá),最終由所有事件的集合來作為視頻片段的主題。該模型以結(jié)合時(shí)序上下文的鏡頭運(yùn)動(dòng)對象和環(huán)境背景組成的事件組作為標(biāo)注結(jié)果。該標(biāo)注模型較好地代表了鏡頭的語義內(nèi)涵,提高了視頻語義表達(dá)的準(zhǔn)確度。(3)提出一種基于半監(jiān)督聚類的視頻標(biāo)注新方法。以鏡頭事件為單位,用事件組來標(biāo)注視頻。為了降低視頻標(biāo)注對已標(biāo)注樣本的依賴,利用半監(jiān)督學(xué)習(xí)思想構(gòu)造半監(jiān)督K-means聚類算法,優(yōu)化目標(biāo)函數(shù),使得最終的聚類結(jié)果既體現(xiàn)類間的低耦合及類內(nèi)的高聚合,又體現(xiàn)類內(nèi)局部的數(shù)據(jù)分布密度。該算法實(shí)現(xiàn)了諸如視頻等多屬性異構(gòu)數(shù)據(jù)的聚類,提高了視頻標(biāo)注的準(zhǔn)確度。(4)提出一種基于上下文的多核學(xué)習(xí)視頻分類新方法。以傳統(tǒng)的詞袋模型為基礎(chǔ),根據(jù)相鄰鏡頭關(guān)鍵幀之間具有相關(guān)性的特點(diǎn)提出了一種用于視頻場景分類的模型。首先將視頻片段進(jìn)行分割,提取關(guān)鍵幀,對關(guān)鍵幀圖像歸一化。接著將關(guān)鍵幀圖像作為圖像塊以時(shí)序關(guān)系合成新圖像,提取新圖像的SIFT特征及HSV顏色特征,并將圖像的SIFT特征及HSV顏色特征數(shù)據(jù)映射到希爾伯特空間。通過多核學(xué)習(xí),選取合適的核函數(shù)組對每個(gè)圖像進(jìn)行訓(xùn)練,最終得到分類模型,得到較好的分類效果。上述研究成果可廣泛應(yīng)用于視頻分類、視頻索引、視頻檢索、視頻內(nèi)容理解、視頻數(shù)據(jù)管理等諸多領(lǐng)域,具有重要的理論意義和較高的應(yīng)用價(jià)值。
[Abstract]:With the rapid development of digital media technology, communication technology and network technology, the number of digital media information represented by video is expanding rapidly. Short video is a kind of video data with a lot of content. How to find effective information in a large amount of short video data has always been a problem of concern to users, resulting in video indexing, video retrieval and other related applications. Video tagging is the core of these applications. At present, video tagging has become a hot research topic in the field of digital media applications and computer vision. From the semantic point of view, video can be divided into several semantic units. Different semantic units have different semantic connotations and can realize semantic annotation at each semantic level. Based on the in-depth analysis of the video structure, the video segment is segmented to form different semantic units, and the short video is annotated in the shot semantic layer and scene semantic layer. The main achievements and innovations of this paper are as follows: (1) combining the global and local features of video frames, a novel shot edge detection method combining video dynamic texture and SIFT features is proposed. In this method, two adjacent frames are partitioned evenly, and the average gradient of each image block in the frame is calculated in RGB color space. The video dynamic texture is formed by the average gradient of all image blocks. The dynamic texture of adjacent frames is compared and the shot change is judged by matching the SIFT features of adjacent frames. This algorithm can detect the shot edge of different types of video data with high accuracy. (2) A video semantic annotation model based on shot events is proposed. Based on the analysis of the video structure, the background color features of the moving object and the key frame of the shot are extracted to express the event of a shot, which extends to the expression of the scene event. Ultimately, the collection of all events is the subject of a video clip. The model takes the event group composed of the shot moving object and the environment background as the annotation result. The annotation model represents the semantic connotation of shot and improves the accuracy of video semantic expression. (3) A new method of video annotation based on semi-supervised clustering is proposed. In the unit of shot event, the video is annotated with event group. In order to reduce the dependence of video tagging on labeled samples, semi-supervised K-means clustering algorithm is constructed by semi-supervised learning idea, and the objective function is optimized, so that the final clustering results can not only reflect the low coupling between classes and high aggregation within classes. It also reflects the local data distribution density in the class. This algorithm implements the clustering of multi-attribute heterogeneous data such as video, and improves the accuracy of video tagging. (4) A new context-based multi-core learning video classification method is proposed. Based on the traditional word bag model, a video scene classification model is proposed according to the correlation between the adjacent shot key frames. Firstly, the video segment is segmented, the key frame is extracted, and the key frame image is normalized. Then the key frame image is used as the image block to synthesize the new image with temporal relation, and the SIFT feature and HSV color feature of the new image are extracted, and the SIFT feature and HSV color feature data of the image are mapped to Hilbert space. Through multi-kernel learning, the appropriate kernel function groups are selected to train each image, and finally the classification model is obtained, and a better classification effect is obtained. These research results can be widely used in many fields such as video classification, video indexing, video retrieval, video content understanding, video data management and so on, which have important theoretical significance and high application value.
【學(xué)位授予單位】：上海大學(xué)
【學(xué)位級別】：博士
【學(xué)位授予年份】：2016
【分類號(hào)】：TP391.41

【相似文獻(xiàn)】

相關(guān)期刊論文 前10條

1 陸懿;陳光夢;畢宏杰;董棟;;改進(jìn)的自然動(dòng)態(tài)紋理綜合算法[J];計(jì)算機(jī)工程與設(shè)計(jì);2008年14期

2 姚偉光;王贏;許存祿;;將局部二進(jìn)制模式應(yīng)用于動(dòng)態(tài)紋理識(shí)別的新方法[J];微計(jì)算機(jī)信息;2010年09期

3 陳昌紅;趙恒;胡海虹;梁繼民;;基于改進(jìn)動(dòng)態(tài)紋理模型的人體運(yùn)動(dòng)分析[J];模式識(shí)別與人工智能;2010年02期

4 陳青;朱俊宇;唐朝暉;劉金平;桂衛(wèi)華;;動(dòng)態(tài)紋理建模在硫浮選工況的識(shí)別分析[J];計(jì)算機(jī)與應(yīng)用化學(xué);2013年10期

5 邵婧;王冠香;郭蔚;;基于視頻動(dòng)態(tài)紋理的火災(zāi)檢測[J];中國圖象圖形學(xué)報(bào);2013年06期

6 陳紅倩;陳誼;曹健;劉鸝;;基于動(dòng)態(tài)紋理技術(shù)的實(shí)時(shí)森林繪制[J];計(jì)算機(jī)仿真;2012年06期

7 何莎;費(fèi)樹岷;;動(dòng)態(tài)紋理背景的建模[J];計(jì)算機(jī)應(yīng)用;2009年S2期

8 鄒運(yùn)蘭;王仁芳;;基于多重紋理和動(dòng)態(tài)紋理技術(shù)的實(shí)時(shí)水面模擬[J];浙江萬里學(xué)院學(xué)報(bào);2010年06期

9 陳紅倩;李鳳霞;黃天羽;戰(zhàn)守義;;一種基于動(dòng)態(tài)紋理的運(yùn)動(dòng)場景可視化方法[J];北京理工大學(xué)學(xué)報(bào);2009年06期

10 于鑫;韓勇;陳戈;;基于動(dòng)態(tài)紋理和粒子系統(tǒng)的火焰效果模擬[J];信息與電腦(理論版);2009年11期

相關(guān)會(huì)議論文 前1條

1 陸懿;陳光夢;;一種改進(jìn)的彩色動(dòng)態(tài)紋理綜合算法[A];中國儀器儀表學(xué)會(huì)第九屆青年學(xué)術(shù)會(huì)議論文集[C];2007年

相關(guān)博士學(xué)位論文 前3條

1 王勇;基于混沌特征向量的動(dòng)態(tài)紋理識(shí)別[D];上海交通大學(xué);2014年

2 彭太樂;基于鏡頭及場景上下文的短視頻標(biāo)注方法研究[D];上海大學(xué);2016年

3 周丙寅;張量分解及其在動(dòng)態(tài)紋理中的應(yīng)用[D];河北師范大學(xué);2012年

相關(guān)碩士學(xué)位論文 前10條

1 陸懿;一種改進(jìn)的基于非線性模型的動(dòng)態(tài)紋理識(shí)別算法[D];復(fù)旦大學(xué);2008年

2 徐磊磊;動(dòng)態(tài)紋理性質(zhì)及其模擬算法研究[D];華中科技大學(xué);2007年

3 姚偉光;基于局部二進(jìn)制運(yùn)動(dòng)模式的動(dòng)態(tài)紋理描述新方法[D];蘭州大學(xué);2009年

4 周文玲;增強(qiáng)現(xiàn)實(shí)中動(dòng)態(tài)紋理的識(shí)別與重建技術(shù)研究[D];華東師范大學(xué);2011年

5 劉霞;自然景物模擬的動(dòng)態(tài)紋理研究與實(shí)現(xiàn)[D];國防科學(xué)技術(shù)大學(xué);2005年

6 丁悅;基于數(shù)據(jù)驅(qū)動(dòng)的馬爾柯夫鏈蒙特卡洛模型的動(dòng)態(tài)紋理分析[D];南京理工大學(xué);2007年

7 曹壽剛;基于李群論和動(dòng)態(tài)紋理的視頻分類技術(shù)研究[D];華中科技大學(xué);2013年

8 高平;基于擴(kuò)展統(tǒng)計(jì)地形特征的動(dòng)態(tài)紋理識(shí)別研究[D];蘭州大學(xué);2009年

9 施濵;基于時(shí)空方向能量的動(dòng)態(tài)紋理研究[D];上海交通大學(xué);2012年

10 張茜;基于動(dòng)態(tài)紋理的流水效果合成技術(shù)研究[D];山東大學(xué);2006年

，

本文編號(hào)：2174386