Upload
srikanth-goud-nagelli
View
219
Download
0
Embed Size (px)
Citation preview
8/11/2019 2.Image Geometry
1/31
8/11/2019 2.Image Geometry
2/31
World everything is in 3D [X,Y,Z]
Lose Z when projected on to 2D detector Methods to recover Z is called as SHAPE
from X where X stands for Stereo, Shaping,
Focus, Defocus etc ..
8/11/2019 2.Image Geometry
3/31
Some Basic Transformation
TranslationScalingRotation
Perspective Transformation
8/11/2019 2.Image Geometry
4/31
Translation
Let (X1,Y1,Z1) displaced to new location givenby
8/11/2019 2.Image Geometry
5/31
8/11/2019 2.Image Geometry
6/31
Scaling
8/11/2019 2.Image Geometry
7/31
Rotation
Rotation around Z theta/ gamma Rotation around X Phi / alpha Rotation around Y beta
8/11/2019 2.Image Geometry
8/31
8/11/2019 2.Image Geometry
9/31
Rotation of a point around Z axis
8/11/2019 2.Image Geometry
10/31
Angles measured counter clockwise
8/11/2019 2.Image Geometry
11/31
Clock wise direction
8/11/2019 2.Image Geometry
12/31
Can you try the ROTATION MATRIX for X and Ydirections [ consider clock wise]
8/11/2019 2.Image Geometry
13/31
8/11/2019 2.Image Geometry
14/31
8/11/2019 2.Image Geometry
15/31
8/11/2019 2.Image Geometry
16/31
8/11/2019 2.Image Geometry
17/31
8/11/2019 2.Image Geometry
18/31
8/11/2019 2.Image Geometry
19/31
8/11/2019 2.Image Geometry
20/31
Perspective Transformation
Also known as Imaging Transformationprojects a 3D point onto a 2D plane.
Here X,Y,Z represent the world coordinatesand x,y,z represent the camera coordinates.
8/11/2019 2.Image Geometry
21/31
X,Y,Z maps to x, y, by considering the two
equivalent triangles
8/11/2019 2.Image Geometry
22/31
If origin moved to the image plane
8/11/2019 2.Image Geometry
23/31
To represent the camera coordinates in a
Matrix form as in case of Translation, Rotationetc.. we need define perspective matrix ,which is accomplished by converting theworld coordinates (Cartesian) to homogenous
world coordinates
8/11/2019 2.Image Geometry
24/31
We define Perspective Transformation
8/11/2019 2.Image Geometry
25/31
8/11/2019 2.Image Geometry
26/31
Cartesian or Euclidean space parallel linesDONT intersect.
in projective space the tracks in the picturebecomes narrower while it moves far awayfrom eyes. Finally, the two parallel rails meetat the horizon, which is a point at infinity.
8/11/2019 2.Image Geometry
27/31
The Cartesian coordinates of a 2D point can be
expressed as ( x, y). What if this point goes far away to infinity? The point at infinity would be (,), and it
becomes meaningless in Euclidean space. The parallel lines should meet at infinity in
projective space, but cannot do in Euclidean
space. Mathematicians have discovered a way to solve
this issue.
8/11/2019 2.Image Geometry
28/31
Homogeneous coordinates,
introduced by August Ferdinand Mbius,make calculations of graphics and geometrypossible in projective space.
Homogeneous coordinates are a way ofrepresenting N-dimensional coordinates withN+1 numbers.
8/11/2019 2.Image Geometry
29/31
To make 2D Homogeneous coordinates, simplyadd an additional variable, w, into existingcoordinates.
Therefore, a point in Cartesian coordinates, (X, Y)becomes (x, y, w) in Homogeneous coordinates. And X and Y in Cartesian are re-expressed with x,
y and w in Homogeneous as; X = x/w Y = y/w
8/11/2019 2.Image Geometry
30/31
8/11/2019 2.Image Geometry
31/31