Bottle Unwrap
The Bottle Unwrap module transforms an image that is wrapped around a bottle surface into a flat surface. This module
is often used as a precursor to an OCR module in order to correct for curve distortion when imaging a curved
surface. For example, the label placed on a cylindrical container can be unwrapped in order to recognize the
words, images, etc. on that surface in an easier planar way.
Interface
Instructions
1. Warp - Adjust the radius of the wrap such that the image appears to flatten out. Numbers larger than the image
width will shrink or compress the image whilst numbers smaller than the current image width will spread out the
image by expanding the image extremes.
2. Z Rotation - If the image is slightly tilted you might want to rotate the image into a more horizontal/vertical
orientation. This will ensure that the bottle unwrap will operate on an aligned image.
3. Y Rotation - If the image is tilted forward or backwards (as if the camera is pointed up or down towards the
label) you might need to adjust the Y Rotation which will rotate the image accordingly.
4. X - If the label in the image is not in the image center use the X transform to move the transform center
into the center of the label. This will ensure that the image unwraps from the correct center point. If your object
is not in the image center and the unwrap is applied you will notice more distortion on one side of the object. Move the
X center until the image appears equally transformed on the left and right sides.
5. Y - Similar to the X transform, the Y transform will move the image center up or down in order to correctly transform
the image.
6. Pixel Resample - specifies that the image will be transformed on a per pixel basis. This is a faster technique and
will make the image appear more crisp but can suffer from being too crisp in that edges may appear too sharp.
7. Bilinear Resample - provides smoother results than the Pixel method but is a slower transform.
8. Display Grid Size - the goal of this module is to align the image into a flat rectilinear surface. In order to
help you tweak the parameters you can enable the Display Grid size to superimpose a grid structure on top of the image
in order to better compare image lines with actual horizontal and vertical lines. Using the grid as a comparison
you can change the values of the transforms in order to achieve the alignment goal.
Example
 Click here
to load a robofile that contains the example transform below. Note that this example includes the
image below and also uses the Radial Distortion module to undistort radial effects due to the camera lens.
| Source |
 |
| Transformed |
 |
See Also
Radial Distortion
|
