Computer Vision

1
Computer Vision

• No. 1
• What is the Computer Vision?

2
Instructor

• Katsushi Ikeuchi
• Pointers
• 03-5452-6242
• cvl-sec_at_iis.u-tokyo.ac.jp
• 4-6-1 Komaba Meguro-ku
• http//www.cvl.iis.u-tokyo.ac.jp

3
Evaluation

• attendance 50
• report 50

4
Schedule

• Shape-from-X
• Analysis of line-drawing
• Shape-from-shading
• Binocular stereo
• Interpretation
• Interpolation
• Representation
• Special topics
• Modeling from reality

5
Katsu Ikeuchi
6
Demonstration Videos
7
Photometric Stereo (1980)

• Brightness difference -gt 3D shape
• 3D shape -gt 3D Pose determination
• 3DPose -gt Grasping

8
Bin Picking
9
Assembly Plan from Observation (1990)
10
Recent Result Assembly plan from observation
11
(No Transcript)
12
Learning Human Dance
13
Motion Capture Data
14
Robot Dancing
15
Modeling Cultural Heritage
16
Virtual City
17
Virtual City
Vehicle
Pedestrian
Speed10km/h
Vehicle
Near Yoyogi park
18
Computer Vision (CV)

• To make a computer to recognize the 3D world as
  we do
• To generate 3D representations from 2D images

19
CV and related areas
Image Understanding (AI)
Pattern Recognition (Mathematical theories)
Image Processing (Signal processing)
20
CV and related areas
Image Understanding (AI)
Pattern Recognition (Mathematical theories)
Image Processing (Signal processing)
21
Image Processing

• To get better images 2D-to-2D

22
CV and related areas
Image Understanding (AI)
Pattern Recognition (Mathematical theories)
Image Processing (Signal processing)
23
Pattern Recognition

• Decision making mathematical theories

24
CV and related areas
Image Understanding (AI)
Pattern Recognition (Mathematical theories)
Image Processing (Signal processing)
25
Image Understanding

• Scene description

26
Why difficult ?

• A lot of data
• Ambiguity
• Projection of a 3D world to a 2D image
• Many factors to influence the image
• Illumination condition
• Object shape
• Camera characteristics

27
Image
Foggy golden triangle in Pittsburgh
28
But
29
A lot of data

• Landsat image
• 1scene 3300 x 2300 x 4 30000000 bytes
• 200 scenes/ day
• Color TV image
• 512 x 512 x 3 x 30 25000000 bytes/sec

30
Why difficult ?

• A lot of data
• Ambiguity
• Projection of a 3D world to a 2D image
• Many factors to influence the image
• Illumination condition
• Object shape
• Camera characteristics

31
Illusion due to the projection
32
Why difficult ?

• A lot of data
• Ambiguity
• Projection of a 3D world to a 2D image
• Many factors to influence to the image
• Illumination condition
• Object shape
• Camera characteristics

33
Image

• A image is a matrix of pixels
• Each pixel
• brightness
• Color
• Distance

34
Inside and Outside (Gestalt)
35
Common sense

• To formulate the common sense
• ? research topics

36
Current issues

• A lot of data
• Computational sensor
• Vision board
• Ambiguity
• Projective geometry
• constraints
• Many factors
• Physics-based vision

37
Application areas
38
Application areas
39
What is Computer Vision?

• Vision is an information processing task that
  constructs efficient symbolic descriptions of the
  world from images. (Marr)
• Vision is inverse graphics.
• Vision is looks easy, but is difficult.
• Vision is difficult, but is fun. (Kanade)
• Vision is an engineering science to create
• an alternative of human visual systems on
  computers (Ikeuchi)

40
References

• Journals
• Inter. J. Computer Vision
• IEEE Trans. Pattern Analysis and Machine
  Intelligence
• IEICE D-2
• IPSJ Trans CVIM
• International conferences
• Inter. Conf. Computer Vision (ICCV)
• Computer Vision and Pattern Recognition (CVPR)
• Asian Conf. Computer Vision (ACCV)
• Special interest groups
• IPSJ CVIM
• IEICE PRMU

41
Schedule (April-May)
4/12 Introduction 4/19 Line drawing 4/26
Perspective projection 5/3 Holiday 5/10 Shape
from Shading 5/17 Color
Dr. Miyazaki 5/24 Stereo1 5/31 Stereo2
Dr. Vanno and Dr. Ogawara
42
Schedule (June-July)
6/7 Motion analysis 6/14 No class 6/21 EPI, IBR
MBR Dr. Ono 6/28 Interpolation 7/5
Object representation1 Dr.Takamatsu 7/12
Object representation2