Skip to content

Normalize

Jump to bottom
Adrian Quintana edited this page Dec 11, 2017 · 1 revision

Normalize

Purpose

This utility allows you to normalize a set of images or volumes. The header of the images is not taken into account to normalize them, i.e., they are read as if no header were present.

Each image is normalized independently according to any of the available methods. Except for theOldXmipp method, the rest DO need a background over which compute some statistics. Some small comments on the normalizing methods:

  • [[OldXmipp]] Not bad, but no positivity constraints can be applied on reconstruction projections. Depends on the CTF. (Check here for details)
  • Near_OldXmipp Very similar to OldXmipp. (Check here for details)
  • [[NewXmipp]] Positivity constraints can be applied on reconstruction projections. Depends on the CTF. (Check here for details)
  • Michael Different projections are not guaranteed to have the same moments. Doesn't depend on the CTF. (Check here for details)
  • Random the projections are randomly normalized. This is used for simulations.. (Check here for details)
  • Ramp Similar to NewXmipp, but with removal of ramp density by subtracting a least-squares plane through the background pixels. (Check here for details)
  • Tomography Similar to OldXmipp but with an extra division by the cos(tilt). The mean and variance are measured only in the region common to the whole tilt series. (Check here for details). It is assumed that the tilt axis is vertical. The images must have the tilt angle in the header.
  • Tomography0 Similar to the previous one but the mean and variance of the image at 0 degrees are used instead of the mean and variance of the image itself. This is useful for images in which a lot of new material enters in the field of view (Check here for details)

If the program is applied to a volume, then the output volume will have 0 mean and standard deviation 1 independently of the normalizing method.

Usage

Parameters

  • `` Input image

  • `` Output image. If no output file is given, then the input one is rewritten.

  • `` Input selfile

  • `` Output extension. If no output extension is given, then the input ones are rewritten

  • `` Do not produce on screen information

  • -method [mthNewXmipp] = Valid methods areOldXmipp, Near_OldXmipp, NewXmipp, Michael, Random, Ramp, Tomography and Tomography0.

  • `` It is needed by all normalization methods exceptOldXmipp. You may use other masks by supplying any of the mask parameters.

    • `` Frame background of r pixels
    • `` circular background outside radius=r

  • `` If the normalization method is random, then the images are randomly transformed using y=ax+b where a and b are randomly picked from a uniform distribution given by [a0,aF] and [b0,bF]

  • `` After normalizing the image, a mask is applied that lets pass only the common part in the whole tilt series.

  • `` Activate the normalization for volumes mode. It sets the average to 0 and the standard deviation to 1

Examples and notes

Normalize a set of images with theOldXmipp method, overwriting input images


$ normalize -i set_images.sel -method OldXmipp

Normalize usingNewXmipp method, for images of 64x64, for instance


$ normalize -i set_images.sel -background circle 29

Normalize usingNewXmipp method, for images of 64x64 using a crown mask


$ normalize -i set_images.sel -mask crown 29 32

Normalize a volume to have zero mean and unit variance


$ normalize -i volume.vol

--Main.AlfredoSolano - 16 Jan 2007

Clone this wiki locally