莊永裕臺灣大學：資訊工程學研究所潘振銘Pan, Chen-MingChen-MingPan2007-11-262018-07-052007-11-262018-07-052007http://ntur.lib.ntu.edu.tw//handle/246246/53886Rushes are the raw materials used to produce a video. They often contain redundant and repetitive contents. Rushes summarization aims to provide a quick overview for a rushes video. As part of TRECVID 2007, NIST initiates a rushes summarization task. This paper reports on the design of NTU rushes summarization system. Our system consists of three components, shot segmentation, redundant shot detection and summary creation. To tackle the bulky rushes, we focus on efficient but effective feature representations (i.e. , local color histograms, compressed-domain motion vectors, etc.) and summarization methods. Besides, we proposed a novel approach to detect clapper shots which are not only relevant to concise summarizes but also essential for indexing numerous camera takes in the rushes. Even practically efficient and requiring only 40% of the video time for computation, the proposed system achieved satisfying results in TRECIVD 2007 rushes summarization task.1 Introduction 1 1.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1.2 TRECVID Task . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 1.2.1 TRECVID . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 1.2.2 Rushes Summarization . . . . . . . . . . . . . . . . . . . . . . . . . . 3 1.3 Organization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 2 Related Work 5 2.1 Shot Boundary Detection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 2.2 Video Summarization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 3 System Overview 9 4 Shot Segmentation 11 4.1 Shot Cut Detection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 4.2 Sub-shot Segmentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 5 Redundant Shot Removal 17 5.1 Junk Shot Detection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 5.1.1 Single Color Detection . . . . . . . . . . . . . . . . . . . . . . . . . . 17 5.1.2 Color Bar Detection . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 5.1.3 Frame Adjustment . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 5.1.4 Calpper Board Detection . . . . . . . . . . . . . . . . . . . . . . . . . 22 5.2 Duplicate Shot Detection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 6 Summary Creation 35 6.1 Segment Selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 6.2 Summary Creation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 v 7 Results and Discussion 39 7.1 TRECVID Result . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39 7.2 Improved Result . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40 8 Conclusion and Future Work 433467046 bytesapplication/pdfen-US影片摘要影片瀏覽Video SummariationVideo SkimRushesthesishttp://ntur.lib.ntu.edu.tw/bitstream/246246/53886/1/ntu-96-R94922014-1.pdf