Cropping Data

You have the option to crop the data but if you do, this must be done after running the normalization. The algorithm only cropped the normalized sample and ob data

  • the 4 corners of the region of interest (ROI)
  • the top left corner coordinates, width and height of the ROI

let’s use the first method and let’s pretend the ROI is defined by

  • x0 = 5
  • y0 = 5
  • x1 = 200
  • y1 = 250
>>> my_crop_roi = ROI(x0=5, y0=5, x1=200, y1=250)
>>> o_norm.crop(roi=my_crop_roi)