import arr,geom
#---------------------------------------------------------------------
# Resample 
# resamples 20x20 mm of originalImage - the origin is defined by the reference identified by referenceName
# the resulting image is of size 60 x 60 pixels and set to image imageName
#---------------------------------------------------------------------
def Resample(originalImage,referenceName,imageName):

  sub_x = 20.0 # section of original image in millimeters -
  sub_y = 20.0 # dimensions given by the reference system
  res_x = 60 # size of resampled image
  res_y = 60 # in pixels
  # get pointer to Reference tool
  t = GetTool(referenceName)
  ref = geom.m33f()
  ref.data = t.childRefSys
  if ( t.result['Status'] == 1 ):  # reference system is valid
     orgim = GetImageMatr(originalImage) # get image data
     # prepare resampling
     p = arr.gridPoints(res_x,res_y,1) # create regular grid matrix of size res_x by res_y, from (0,0)
     p = arr.toFix(p*geom.scal(sub_x/res_x,sub_y/res_y)*~ref) # overlay grid points on original image, using toFix for efficiency
     # resample image
     subim = arr.bilInterpolate(orgim,p) # extracts samples from orgim at positions given by p
     # set the resampled image to imageName 
     SetImageMatr(imageName,arr.toUint8(subim))

#--------------------------------------------------------------------

Such resampling is especially interesting and useful when one combines resampling with the powerful reference systems.

In the Scrabble example a letter is placed under the camera. The size and orientation of this letter may vary.

By establishing a reference system that finds the letter and by resampling one can produce a clean letter image

This image has the same size and orientation no matter how the camera or letter is positioned.