Copying, Reordering and Regridding Images

• The task imsub allows you to extract out some region of one image and copy it to a new image. For example, let us copy every other pixel from the central 20 by 20 arcsec from the first 20 planes of a cube into a new file. Naturally, imsub can also be used to make an identical copy of your entire image.

  

• The task imcat allows you to concatenate images along the third dimension. Thus you can build cubes from 2-D images, and bigger cubes from smaller cubes. One use, for example, is that you might like to put dissimilar images (e.g. 2-D images from different spectral lines) into the one cube and make a Post-Script plot of all of them together on one plot (each plane in its own sub-plot) with cgdisp .

  When concatenating dissimilar images, you will probably want to set options=relax. This tells imcat not to be concerned if axis descriptors differ. However, the image dimensions must still be the same.

  

• The task reorder enables you to reorder the axes of an image. For example, to reorder a cube in xyv order to vxy order and flip the direction of the velocity axis you would do the following:

  

• Similar to reorder is imframe . It too can reorder the axes , but generally we recommend you use reorder for that application as its inputs are much simpler than those of imframe . imframe is useful if you want embed an image inside a larger one and pad the border with zeros.

  In the example, we take an image of some size and pad it out to be 512 by 512 pixels. Note that the frame keyword which specifies the output image size in relative pixels is in the non-standard order of xmin,xmax,ymin,ymax compared to the region keyword which expects xmin,ymin,xmax,ymax.

  

• When comparing images, it is often necessary to have them on the same grid so that tasks such as cgdisp can overlay them graphically, or so that maths (see below) can evaluate some expression involving the two images.

  The task regrid enables you to take an image, and regrid it via spline fitting and resampling on a new grid. This task can regrid any combination of the the first three axes of an image. The new grid can be specified in one of three ways. The first way (and generally easiest) is to take the header of another (template) image and make the image of interest look like the template image. The second way requires you to specify the output header axis descriptors explicitly with the keyword desc. You have to input the reference value, reference pixel, pixel increment and number of pixels for each axis you want to regrid -- pretty tedious stuff. The third way is to specify, via the keyword range, the start, end and increment for each axis you wish to regrid.

  regrid currently does not handle non-linear axes correctly. It assumes that the axis is linear as given by the header axis descriptors.

  In this simple example, we regrid the first two axes of an image via a template image.