1. Nov. 2015 Visualization with 3D CG
3D Digitization
Masaki Hayashi

2. Photogrammetry (break)
Today’s contents
Laser scanning
Photogrammetry (break)
Agisoft “Photoscan”

3. Laser scanning

4. Range: Very small object (eg. tooth), Middle (eg.
Laser scanning
Range: Very small object (eg. tooth), Middle (eg.
Statue), Very large (eg. ruin, city landscape)
+ Aerial (Huge area)
Output: 3D point cloud, basically.
x, y, z coordinates
+ I(intensity) + r, g, b(color)

5. Short-range ( < 1 m ) Laser triangulation Structured light
How does 3D Scanner work?
Short-range ( < 1 m ) Laser triangulation Structured light
Middle-range (10m) Travel Of Light (TOL): Phase shift
Long-range ( > 100m ) Travel Of Light (TOL): Pulse

6. Short-range ( < 1 m ) Laser triangulation Structured light Laser
Lens
Sensor
Triangulation
Known
(L)
Object
Measured
Measured
(β)
(α)
Structured light
(D)
(more accurate)
Linear patterns
Sensor
Calculated by 3 values
Lens
L sinβ
sin (α + β)
D =
Object

7. Long-range ( 100 m ) TOL (Travel Of Light) Measure the TOL Laser
c (m/sec) = 299,792,458
Sensor
Lens
Object
Out
Out In In
TOL
TOL (phase)
Pulse modulation
Phase modulation
(more accurate)

8. Laser scanning Multiple measurements: Compensation of the shadow
Planning:
Locations
Angle of view
Resolution

9. Laser scanning
Error:
Beam expansion Spot diameter is bigger when range is bigger e.g. 12mm at 5m  97mm at 500m
Depending on material Diffuse on the surface (clod, solid, reflective…)
Angle of the beam when the beam strikes the surface

10. Laser scanning Example video http://youtu.be/YpcGmh85Hes
Minolta-Konica Vivid 9i

11. Laser scanning
Some scenes

12. Photogrammetry

13. Range: Almost same as scanner Small object (eg. vase), Middle (eg.
Photogrammetry
Range: Almost same as scanner Small object (eg. vase), Middle (eg.
Statue), Very large (eg. ruin, city landscape)
+ Aerial (Huge area)
Output: 3D point cloud + Color
x, y, z + r, g, b
Good point: Texture color is more realistic.
Bad point: More errors.

14. How does photogrammetry work?
Camera
Camera
3D position can be calculated if the camera parameters are known
by triangulation
Object
Camera parameter:
x, y, z, rx, ry, rz, AngleOfView
Known points
(Feature point)
All the calculation is automatically done at a time
Don’t get the software confused to get feature points.

15. Photogrammetry What the software does first is: error
Uses image processing algorythm to get as much feature points on a given images.
Searches the corresponding feature points automatically.
Generally, it produces a lot of errors.

16. Recent tech Better Reality : Thorskan https://youtu.be/S06y8Va68XE

17. Recent tech