[Insight-users] questions about the 3D image registration using OnePlusOneEvolutionary Optimizer with mattesMI metric

[Karthik Krishnan]

LiuYuanJie wrote:

> Hi,everyone. I want to implement a function which can do the
> registration of 2 3d images.
>
> I use oneplusoneevolutionary optimizer and use mattes metric.
>
> Maybe I set some parameters wrong so that i can't get a perfect
> result. (In fact,when I choose 2 same image and firstly do some
> transforms and rotations on the first image and do registration with
> the second image.The result is still similar to my initial parameters
> instead of 0.
>
> Someone told me that maybe the cause is that the StepLength of the
> optimizer is too small. But it seems that the oneplusoneevolutionary
> optimizer does not have a method which sets value to steplength or I
> don't know how to use.
>
The equivalent set of parameters here are the shrink and growth factors.
The optimizer throws random samples in a search radius and tries to hone
in on a solution. Please take a look at

"Evaluation of 2D/3D bias correction with 1+1ES-optimization" Martin
Styner, Prof. Dr. G. Gerig (IKT, BIWI, ETH Zuerich), TR-197
http://www.cs.unc.edu/~styner/docs/StynerTR97.pdf <http://www.cs.unc.edu/%7Estyner/docs/StynerTR97.pdf>

> So I wonder someone can tell me how to set the parameter correctly so
> that i can get a much better result.
>
> Listed below are my codes of optimizer and metric:
>
> optimizer->SetNormalVariateGenerator(generator);
> optimizer->Initialize(10);
> optimizer->SetEpsilon(0.5);
> optimizer->SetGrowthFactor(-100);
> optimizer->SetMaximumIteration(4000);
>
It does not make sense for the growth factor to be -100. The growth
factor is usually slightly greater than 1. A typical value may be 1.01
The shrink factor is by default set to pow( growth_factor, -.25 ).

Please take a look at the reference.

>
>
>
>
>
>
>
> 想 要 一 个 快 n 倍 的 免 费 邮箱 吗 ？
> 126 专 业 电 子 邮局 ―― 全 球 领 先 的 中 文 邮 箱 服 商 带 你 进 入
> 极 速 之 旅 <http://www.126.com/help/126fab.htm>
>
>------------------------------------------------------------------------
>
>_______________________________________________
>Insight-users mailing list
>Insight-users at itk.org
>http://www.itk.org/mailman/listinfo/insight-users
>  
>