課程資訊
課程名稱
 三維動態環境感知
3D Dynamic Environment Perception 
開課學期
 101-1 
授課對象
 電機資訊學院  資訊網路與多媒體研究所  
授課教師
 王傑智 
課號
 CSIE7422 
課程識別碼
 922 U4040 
班次
  
學分
 3 
全/半年
 半年 
必/選修
 選修 
上課時間
 星期二5,6,7(12:20~15:10) 
上課地點
 資107 
備註
 限學士班三年級以上
總人數上限：50人 
Ceiba 課程網頁
 http://ceiba.ntu.edu.tw/10113ddep 
課程簡介影片
 
核心能力關聯
 本課程尚未建立核心能力關連
課程大綱
為確保您我的權利,請尊重智慧財產權及不得非法影印
課程概述

FINDING A SOLUTION TO 3D DYNAMIC ENVIRONMENT UNDERSTANDING IS A KEY PREREQUISITE FOR MAKING ROBOTIC SYSTEMS TRULY AUTONOMOUS AND MAKING MANY POTENTIAL APPLICATIONS FEASIBLE. 3D DYNAMIC ENVIRONMENT PERCEPTION IS ONE OF THE MOST IMPORTANT RESEARCH TOPICS IN BOTH ROBOTICS AND COMPUTER VISION. IN THIS COURSE, THE STATE-OF-THE-ART APPROACHES IN 3D DYNAMIC ENVIRONMENT PERCEPTION USING A VARIETY OF SENSORS SUCH AS A MONOCULAR CAMERA, STEREO CAMERAS, LASER SCANNERS AND RGB-D CAMERAS WILL BE DESCRIBED. 

課程目標
 THIS CLASS WILL COVER GEOMETRY-BASED AND LEARNING-BASED METHODS IN 3D DYNAMIC ENVIRONMENT PERCEPTION WITH AN EMPHASIS ON SEGMENTATION AND RECONSTRUCTION OF DYNAMIC SCENES. TOPICS INCLUDE: SENSORS (CAMERAS, LASER SCANNERS AND RGB-D CAMERAS), RECONSTRUCTION OF STATIC SCENES (TRACKING, DATA ASSOCIATION, MULTI-VIEW GEOMETRY, STRUCTURE FROM MOTION, MONOCULAR SIMULTANEOUS LOCALIZATION AND MAPPING), RECONSTRUCTION OF DYNAMIC SCENES (MOTION SEGMENTATION, NON-RIGID STRUCTURE FROM MOTION, MULTI-BODY STRUCTURE FROM MOTION, MONOCULAR SLAMMOT), AND 3D FACE ALIGNMENT AND RECONSTRUCTION. 
課程要求
 You should have taken any robotics- and/or computer vision- related courses. Hands-on experience with software development in Matlab, C or C++ will be helpful for this course. But the most important prerequisite will be creativity and enthusiasm, and a desire to explore.
 
預期每週課後學習時數
  
Office Hours
 每週二 12:00~13:20 
指定閱讀
 To be announced.  
參考書目
  
評量方式
(僅供參考)
    
課程進度
週次
日期
單元主題
Week 1
 9/11  Introduction 
Week 2
 9/18  Representation of a 3-D Moving Scene 
Week 3
 9/25  Image Formation, Image Primitives and Correspondence  
Week 4
 10/02  Image Primitives and Correspondence 
Week 5
 10/09  Reconstruction from Two Calibrated Views 
Week 6
 10/16  Midterm Exam 1 
Week 7
 10/23  Reconstruction from Two Uncalibrated Views 
Week 8
 10/30  Reconstruction from Two Uncalibrated Views, and Multiple View Geometry of Points 
Week 9
 11/06  Final Project Proposal Presentation and Midterm Exam 1 discussion 
Week 10
 11/13  Multiple View Geometry of Points, Structure from Motion, and Range Sensors 
Week 11
 11/20  Non-rigid Structure from Motion 
Week 12
 11/27  Midterm Exam 2 
Week 13
 12/04  NRSFM  
Week 14
 12/11  SLAM and Monocular SLAM 
Week 15
 12/18  SLAMMOT 
Week 16
 12/25  SLAMMOT 2.0 
Week 17
 1/01  National Holiday. No Class. 
Week 18
 1/08  Final Project Presentation