Special features

Previewing

Previewing enables the process list to be applied to a subset of the data. Each loader plugin has a preview parameter that is empty by default (apply processing to all the data). The preview requires a list as input with entries for each data dimension. Each entry in the preview list should be of the form start:stop:step:chunk, where stop, step and chunk are optional (defaults: stop = start + 1, step = 1, chunk = 1) but must be given in that order. For more information see set_preview()

Previewing Examples

The 3-D NxtomoLoader plugin maps the data dimensions (0, 1, 2) to the axis labels (rotation_angle, detector_y, detector_x) respectively.

>>> savu_config
>>> add NxtomoLoader
>>> mod 1.1 [:, mid-5:mid+6, :]     # process the middle 10 sinograms only
>>> mod 1.1 [0:end:2, mid-5:mid+6, :]      # process every other projection
>>> mod 1.1 [0:end:2, mid-5:mid+6, 300:end-300] # crop 300 pixels from the sides of the detector

Sinogram centering

Automatic calculation OR manual input of the centre of rotation are possible in Savu.

Auto-centering

The auto-centering plugin (VoCentering) can be added to a process list before the reconstruction plugin. The value calculated in the centering routine is automatically passed to the reconstruction and will override the centre_of_rotation parameter in the reconstruction plugin. The auto-centering plugin is computationally expensive and should only be applied to previewed data. There are two ways to achieve this:

  1. Apply previewing in the loader plugin to reduce the size of the processed data.

and/or

  1. Apply previewing in VoCentering plugin (this will not reduce the size of the data).

Note

If you have applied previewing in the loader and again in the centering plugin you will be applying previewing to the previewed (reduced size) data.

See Full pipeline with auto-centering

Manual Centering

Ensure the VoCentering algorithm is not in the process list (remove it or turn it off if it is already inside your list). Modify the centre_of_rotation value in the reconstruction plugin, see Apply manual centering. If the manual centering value is approximate you can apply parameter tuning, see Apply parameter tuning to the centre of rotation

Parameter Tuning

If you wish to test a preview reconstruction with a range of values for a parameter, for instance, if the centering is not quite optimal, then you can add different values separated by semi-colons. Each ‘tuned’ parameter will add an extra dimension to the data.

Parameter tuning examples

>>> mod 6.7 85;86;87        # three distinct values
>>> mod 6.7 84:86:0.5;      # a range of values (start:stop:step) with semi-colon at the end
>>> mod 6.6 FBP;CGLS        # values can be strings

See Example 2 and Apply parameter tuning to the centre of rotation.