# Mattes et al. vs Viola and Wells Mutual Information

**URL:** <https://discourse.itk.org/t/mattes-et-al-vs-viola-and-wells-mutual-information/1645>\
**Category:** Algorithms\
**Created:** [March 8, 2019, 9:07pm UTC](https://discourse.itk.org/t/mattes-et-al-vs-viola-and-wells-mutual-information/1645 "2019-03-08T21:07:41Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![ibr](https://discourse.itk.org/user_avatar/discourse.itk.org/ibr/32/468_2.png) [@ibr](https://discourse.itk.org/u/ibr)\
**Post date:** [March 8, 2019, 9:07pm UTC](https://discourse.itk.org/t/mattes-et-al-vs-viola-and-wells-mutual-information/1645/1 "2019-03-08T21:07:41Z")

</div>

Hi,

Can someone please explain the difference between [Mattes et al Mutual Information](https://itk.org/Doxygen/html/classitk_1_1MattesMutualInformationImageToImageMetric.html) MMI and [Viola and Wells Mutual Information](https://itk.org/Doxygen/html/classitk_1_1MutualInformationImageToImageMetric.html) VMI?

From the description in the links, I understand that MMI uses Parzen histogram with a user-defined number of pins ( it is not clear what is the difference between Parzen window and Parzen histogram?). MMI uses 3rd-order BSpline kernel for only the moving image (in Viola and Wells it uses Gaussian kernel for both images).

Thanks