# Multi-level neighborhood-correlation BSpline

**URL:** <https://discourse.itk.org/t/multi-level-neighborhood-correlation-bspline/296>\
**Category:** Algorithms\
**Created:** [October 16, 2017, 7:08pm UTC](https://discourse.itk.org/t/multi-level-neighborhood-correlation-bspline/296 "2017-10-16T19:08:19Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![Ray](https://discourse.itk.org/letter_avatar_proxy/v4/letter/r/e9a140/32.png) [@Ray](https://discourse.itk.org/u/Ray)\
**Post date:** [October 16, 2017, 7:08pm UTC](https://discourse.itk.org/t/multi-level-neighborhood-correlation-bspline/296/1 "2017-10-16T19:08:20Z")

</div>

I’m using neighborhood correlation for registration metric and BSpline for transformation model. The images are 100x100. The neighborhood window size is 10pixel-by-10pixel. The mesh size is 10x10.

If I use 2-level registration, and set the shrink factor to be 2. The images will become 50x50 at the higher level.

Is the neighborhood window size 5pixel-by-5pixel or 10pixel-by-10pixel at the higher level?

Is the mesh size 5x5 or 10x10 at the higher level?

Thanks!  
-Ray