[Insight-users] A question about least_square method !

[Luis Ibanez]

Hi Zhen,


Do you want to use Least Squares in order to fit a function ?

If so, you have multiple options, and they depend on whether your
function is linear or not.




1) You can use the KalmanLinearEstimator
http://www.itk.org/Insight/Doxygen/html/classitk_1_1KalmanLinearEstimator.html
    in order to build the estimation progressively.
    The final output of the KalmanLinearEstimator is
    equivalent to a least-squares fit.


2) If your function is non linear you can fit it
    by using the Levenberg-Marquardt optimizer
http://www.itk.org/Insight/Doxygen/html/classitk_1_1LevenbergMarquardtOptimizer.html



3) If you can express your function as a set of linear
    relationships then you can solve it using SVD.
    In this case you may want to use the VNL classes:

      Insight/Utilities/vxl/core/vnl/algo
                    vnl_svd.h
                    vnl_svd.txx
                    vnl_svd_economy.h
                    vnl_svd_economy.txx



4) You can also use VNL optimizers and the vnl_least_squares function

      Insight/Utilities/vxl/core/vnl/
                 vnl_least_squares_function.h




   Regards,



       Luis



----------------------
zhen zonglei wrote:

> Hi everyone.
>  I am a new user of ITK.
> I  want to konw whether the least_square method have been implemented
> in ITK as a common Numerical method, and which class can provide the
> LS method?
> 
> Thanks!
> 
>