This operation is done to correct of the pixel to pixel variations. A flat image with high signal to noise ratio is necessary. And the relative displacement between the apertures of this flat image and the object image have to be lower than 0.1 pixel.
Firstly, we redefine the apertures to the standard widths with apall.
Secondly, we use apflatten (see Appendix I) to create an image (hereafter, flatresponse) which is exactly equal to one outside the apertures width and around to one (depending on pixel to pixel relative sensibility variations) inside them.
Finally, we divide the object image (after bias subtraction and scattered light subtraction) by the flatresponse. This operation can be done with imarith (load images package).