[Scipy-tickets] [SciPy] #1497: Add Gaussian kernel convolution to interpolate.interp1d and interpolate.interp2d

SciPy Trac scipy-tickets@scipy....
Sun Aug 21 11:00:35 CDT 2011


#1497: Add Gaussian kernel convolution to interpolate.interp1d and
interpolate.interp2d
-------------------------------+--------------------------------------------
 Reporter:  miguel             |       Owner:  somebody   
     Type:  enhancement        |      Status:  needs_info 
 Priority:  normal             |   Milestone:  Unscheduled
Component:  scipy.interpolate  |     Version:  0.9.0      
 Keywords:                     |  
-------------------------------+--------------------------------------------

Comment(by miguel):

 Replying to [comment:2 pv]:
 > As I understand, the simple kernel convolution is straightforward to
 implement only for data on regularly spaced grids. If so, it is not
 suitable for interp1d and interp2d, which must support data on irregularly
 spaced grids.

 It should be possible to use kernel convolution also for non-uniform
 grids. The 1D and 2D convolution formulas for an arbitrary point are given
 at:
 http://en.wikipedia.org/wiki/Lanczos_resampling#Interpolation_formulas ,
 where the kernel function could be provided by the user.

 > However, we could use a new function for working with regularly spaced
 data. In principle `scipy.ndimage.map_coordinates` does kernel
 convolution, but it is not easy to discover or very convenient to use for
 interpolation (because it is was written for image processing purposes).

 It seems that the scipy.ndimage.map_coordinates function only supports
 spline interpolation. There is also a scipy.ndimage.convolve function that
 perhaps could be adapted for interpolation.

-- 
Ticket URL: <http://projects.scipy.org/scipy/ticket/1497#comment:3>
SciPy <http://www.scipy.org>
SciPy is open-source software for mathematics, science, and engineering.


More information about the Scipy-tickets mailing list