|
SUPPORT SITE |
|
|
|
SUPPORT SITE |
|
Path: SCPro/ProSensors
% Generate a camera transformation.
Transforms from the scene frame to the pixel frame. Does not do
depth buffering. The points you give it must be visible to the camera
This outputs the transformed points and the transformation matrix.
types are 'perspective', 'affine', 'infinite' and 'weak perspective'.
With 'infinite' the input points should be unit vectors. They are
unitized internally.
The radial distortion model is xd = x*(1 + k(1)r^2 + k(2)r^4)
where r^2 = x^2 + y^2
Type CameraTransform for a demo. The first three apply the
'perspective', 'affine' and 'weak perspective' transforms to the
vertices of a triangle. The last applies the infinite transform
to a point on the boresight.
--------------------------------------------------------------------------
Form:
[pT, m] = CameraTransform( type, d, p )
or
m = CameraTransform( type, d )
--------------------------------------------------------------------------
------
Inputs
------
type (1,:) Transformation type
d (1,1) Data structure
.t (3,1) Location of camera
.r (3,3) Rotation matrix from scene to camera frame
.o (2,1) Origin of the pixel frame (pixels)
.n (2,1) Dimensions of the pixel array
.f (1,1) Focal length (m)
.s (2,1) Scaling (m/pixel)
.k (1,2) Distortion coefficients
p (3,:) Points to transform
-------
Outputs
-------
pT (2,:) Transformed points or m if p not entered
m (3,3) Transformation matrix
j (1,m) Removed points
--------------------------------------------------------------------------
Reference: E. Trucco and A. Verri, "Introductory Techniques for 3-D
Computer Vision, Prentice-Hall, 1998.
--------------------------------------------------------------------------
Common: Graphics/NewFig Common: Graphics/TitleS Common: Graphics/XLabelS Common: Graphics/YLabelS Common: Graphics/ZLabelS Math: Linear/Unit
Back to the SCPro Module page