Until late 2007, I had never needed to embed images into my postscript, so I considered it "deep magic". In 2007, I volunteered to automate creation of ID cards for a club, so I needed to be able to render supplied images of the members.
I discovered that postscript's image format is not dissimilar from the PPM and PGM standards used by the widely available netpbm utilities. They are in fact similar enough that you can read in ppm and pgm files directly from postscript code. This is probably not useful when writing postscript to be run by a printer, but if (like me) you mostly render your postscript to pdf or image formats, it greatly simplifies adding images to your pages.
This library provides the functions drawpnm_width, drawpnm_height, drawpnm_square, and drawpnm_fit. All functions simply take a string argument giving the location of the image file. The first two functions scale the image to have a width of 1 unit or a height of 1 unit respectively, and put the lower left corner of the image at 0,0. The third function trims the image down on the top and bottom or left and right to make it a square and paints that square in the 0,0 - 1,1 square. The fourth scales the image so that it fits entirely in the unit square, and centers it within that square.
The library will deal with standard pgm (greyscale) and ppm (color) images with no change to the ps code. The third pnm format, pbm, differs slightly from postscript's 1-bit image format. Postscript images have every 8 pixels collated into a byte. PBM images do the same, except that the end of each line is padded with zeroes until a byte boundary is reached. There's probably some way around this, or even a way to handle the "plain" formats, but I didn't think they were worth figuring out - it's easy to convert pbm to pgm (by using the pbmtopgm command, for instance).
I release this code to the public domain - do what you like with it.