2D and 3D Registration

Process: Fusion: 2D Registration:

• Aligns 2D images
• Specify parameters to load
• Eliminates hours of Photoshop
  
  

Steps:

1. Determine registration mode: sequential or single reference (Generate: Control...)
2. Determine surface matching parameters: Rotation, Translation, Sampling, Thresholds (Generate: Surface Match...)
3. Register (Generate: Surface Match...)
4. Transform... (Generate: Transform...)
5. Results saved to "desktop": save to disc

Process: Fusion: 3D Registration:

• Aligns 3D voxel sets
• Larger error-space
• More parameters to control matching

Notes

One of the most common problems with 2D data is ensuring that the images are properly aligned. The datasets are often rotated slightly, making it difficult to get accurate reconstructions of the data. The registrations routines automate the process that people manually perform, elminating hours of playing with Photoshop.

During surface matching, parameters can be specified which control the maximum rotation and translation to apply to the individual images. The sampling levels and thresholds can also be specified to control which pixels are matched to one another. If the dataset is rotated or aligned incorrectly, the starting and stopping ranges can be adjusted to correct the problem.

The results are saved to the workspace, so be sure to save back to your local disc.

Registration can also be performed on 3D datasets, such as 3D datasets generated over time. Voxels are compared instead of pixels. It works in a similar manner to the 2D, though further parameters may be specified to restrict the extra degrees of freedom.

For Demo

• Select the AppendedFiles0 dataset (preloaded)
• Process: Fusion: 2D Registration...
• Generate: Control...
• Generate: Surface Match... press Register
• Generate: Transform...
• Point to transformed data on the workspace

Page 11 of 17