# Better positions of the subjects of registration

**URL:** <https://discourse.itk.org/t/better-positions-of-the-subjects-of-registration/3240>\
**Category:** Algorithms\
**Created:** [June 30, 2020, 10:27am UTC](https://discourse.itk.org/t/better-positions-of-the-subjects-of-registration/3240 "2020-06-30T10:27:19Z")\
**Posts on this page:** 4\
**Page:** 1

<div class="post-metadata">

**Author:** ![Luis\_Carlos\_Carneiro](https://discourse.itk.org/user_avatar/discourse.itk.org/luis_carlos_carneiro/32/1486_2.png) [@Luis\_Carlos\_Carneiro](https://discourse.itk.org/u/Luis_Carlos_Carneiro)\
**Post date:** [June 30, 2020, 10:27am UTC](https://discourse.itk.org/t/better-positions-of-the-subjects-of-registration/3240/1 "2020-06-30T10:27:19Z")

</div>

Dear All,

I am using the following example.

[http://insightsoftwareconsortium.github.io/SimpleITK-Notebooks/Python\_html/60\_Registration\_Introduction.html](http://insightsoftwareconsortium.github.io/SimpleITK-Notebooks/Python_html/60_Registration_Introduction.html)

I have the following doubt. If I want to do the registration of two faces. Are a better (privileged) orientation/position of the faces for a better registration? Or only interest the relative positions of the faces?

Thanks,

Luis Gonçalves

---

<div class="post-metadata">

**Author:** ![zivy](https://discourse.itk.org/user_avatar/discourse.itk.org/zivy/32/1726_2.png) [@zivy](https://discourse.itk.org/u/zivy)\
**Post date:** [June 30, 2020, 12:59pm UTC](https://discourse.itk.org/t/better-positions-of-the-subjects-of-registration/3240/2 "2020-06-30T12:59:39Z")

</div>

Hello @Luis_Carlos_Carneiro,

The absolute pose(orientation+position) of the two is not critical and is most often assumed to be accounted for by the initialization which brings them into sufficient proximity so that the residual relative pose is recovered by registration (the iterative optimization process).

---

<div class="post-metadata">

**Author:** ![Luis\_Carlos\_Carneiro](https://discourse.itk.org/user_avatar/discourse.itk.org/luis_carlos_carneiro/32/1486_2.png) [@Luis\_Carlos\_Carneiro](https://discourse.itk.org/u/Luis_Carlos_Carneiro)\
**Post date:** [June 30, 2020, 6:13pm UTC](https://discourse.itk.org/t/better-positions-of-the-subjects-of-registration/3240/3 "2020-06-30T18:13:28Z")

</div>

```
registration_method.SetShrinkFactorsPerLevel(shrinkFactors = [4,2,1])
registration_method.SetSmoothingSigmasPerLevel(smoothingSigmas=[2,1,0])
registration_method.SmoothingSigmasAreSpecifiedInPhysicalUnitsOn()

```

I suppose that the above lines do not affect direction but defines sub-samples of the image.

---

<div class="post-metadata">

**Author:** ![zivy](https://discourse.itk.org/user_avatar/discourse.itk.org/zivy/32/1726_2.png) [@zivy](https://discourse.itk.org/u/zivy)\
**Post date:** [June 30, 2020, 6:17pm UTC](https://discourse.itk.org/t/better-positions-of-the-subjects-of-registration/3240/4 "2020-06-30T18:17:19Z")

</div>

Yes, these define the multi-resolution pyramid size and smoothing/blurring.