黃肇雄臺灣大學：資訊工程學研究所梁致豪Liang, Chih-HaoChih-HaoLiang2007-11-262018-07-052007-11-262018-07-052004http://ntur.lib.ntu.edu.tw//handle/246246/53827隨著高速網路的出現以及各式數位資訊產品的普及，數位內容越來越容易取得及利用。然而大量的數位內容同時也耗費使用者相當多的時間去管理及佔用大量儲存內容的空間。因此如何有效地利用數位內容的資訊以及建立一套有效率的管理機制是刻不容緩的議題。方便的數位內容管理工具將提高使用者使用數位內容的興趣同時也可節省下大量的時間。針對數位內容本身的特性而發展合適的管理工具是目前常見的方法。 運動節目是最受歡迎的節目之一，精采的比賽片段往往是球迷愛不釋手的珍藏品。大量的球賽影片因而廣泛的流傳在網路上及儲存於數位設施之中。下載影片前若能夠先大致瀏覽影片的摘要將減少使用者下載到非所需影片的機會。如果數位內容管理工具能夠對影片內容做分析，進而整理出較具結構的內容索引，使用者就可以更方便的觀看或儲存這些數位影片。 這篇論文主要探討的是棒球影片的內容分析及應用，最主要的目標是偵測出棒球影片中出現的棒球事件的片段。棒球事件是棒球比賽中最精采的部份，同時也呈現了棒球比賽的流程。We can easily capture and use the digital content since the high-speed network and the digital devices have appeared and get popular. The huge amount of digital content causes the users much of time and many of spaces to deal with them. Therefore, how to effectively use the information contained in the digital content and establish an efficient content management tool is an important issue. Powerful tools will appeal users to use the digital content frequently and help them save much time. The primary step to establish an useful management tool is based on the understanding of the characteristics of digital content. Sports program is one of the most popular programs and the interesting segments always appeal a lot of audiences to collect. A huge number of sports videos spread over the networks and occupy much storage space in the digital devices. The summaries of videos can provide users previews and reduce the burdens of downloading redundant videos. If the digital content management tools can analyze the videos and establish a well-defined video index, users can review and store the videos more effectively. The focus of this thesis is on the analysis and application of baseball videos. We propose some methods to detect meaningful baseball events in the videos. The event-contained segments are the most interesting subjects in the videos, and the events also reflect the progresses of the game.Chapter 1 Introduction 1.1. Motivation 1.2. Literature Survey 1.2.1. Video Classification 1.2.2. Scene Classification 1.2.3. Event Detection 1.2.4. Other Approaches 1.3. Contributions 1.4. Thesis Organization Chapter 2 Preliminaries 2.1. Introduction to Baseball games 2.1.1. Events in Baseball game 2.1.2. Characteristics of the Baseball game 2.2. A Brief Description of the Adopted Color space 2.2.1. The RGB Color space 2.2.2. The HSL/HSI Color spaces Chapter 3 Event Boundary Decision 3.1. Color space Selection 3.1.1. RGB to HSI Transformation 3.1.2. Color Histogram 3.1.3. Color Detection 3.1.4. Dominant Color Decision 3.2. Shot Boundary Detection 3.3. Keyframe Extraction 3.4. Pitch shot Detection 3.4.1. Dominant Color Decision 3.4.2. Pitch shot Verification 3.4.3. Pitch shot Cluster Chapter 4 Event Detection 4.1. Extraction of the Game’s Status 4.1.1. Text Status Extraction 4.1.2. Extraction of Symbol Status 4.2. Event Decision 4.2.1. Definition of Event Model in Baseball Game Chapter 5 Combinations and Applications 5.1. Structural Analysis of Baseball Videos 5.2. Summarization of Baseball Videos Chapter 6 Experimental Results 6.1. Test Data Set 6.2. Event Boundary Decision Results 6.3. Event Detection Results 6.4. Application Results Chapter 7 Conclusions and Future Work 7.1. Conclusions 7.2. Future Work2792818 bytesapplication/pdfen-US棒球意義事件影片eventsemanticvideobaseballthesishttp://ntur.lib.ntu.edu.tw/bitstream/246246/53827/1/ntu-93-R91922014-1.pdf