norm

From VoxBoWiki

norm is a command-line implementation of the spatial normalization routine used by SPM96. Like realign, it's just a straight port from MATLAB to c/c++. norm is quite outdated, and you probably want to be using something else. Many VoxBo users rely on SPM's normalization routine. However, if you really want to use it...

The basic idea is that first you use the -calc option to calculate a mapping from one of your images to a reference image, and that later you use the -apply option to apply that mapping (stored in the param file) to a bunch of images. Prototypically, you might have a high-resolution structural image and a set of lower-resolution functional images that occupy the same space. You might also have a template image somewhere, possibly one of the MNI templates. In this typical case, you would calculate the normalization parameters using the high-resolution structural images. You would then apply those parameters to that subject's structural and functional images.

norm will try to guess the origin of any image for which none is specified in the header. It will also guess the bounding box. It will do a reasonable job for both if your image is in the same orientation as the standard reference volumes (x is left/right, y is anterior/posterior, and z is inferior/superior). Otherwise it might do something strange. It also depends on your structural and functional images having the same origin, so that the parameters calculated from one can be applied to the other. See the documentation for setorigin for some ideas on how to arrange this.

Here's the help for norm: