Flat Fields

Flat field exposures are used to remove the effects of vignetting and dust from normal light frame exposures, and to correct for fixed pixel-to-pixel variations. It is often instructive to construct a high signal-to-noise flat field frame and examine it closely for the presence of any inherent stripes, banding or other large or small scale variations.

A series of five flat field exposures was made using a camera lens attached to the front of the CCD camera, illuminated by a white-light LED light box. The central part of each flat field frame had a level of about 40000 ADU. An average flat field frame was made by summing all the exposures and dividing by five. This average flat field was then debiased. There was some vignetting evident away from the centre of the frame. This was 'removed' by fitting a fourth-order polynomial surface to the flat field frame and then dividing the flat field frame by this surface.

Flat field structure

A section of 600 by 400 pixels from the middle of the flattened frame is shown below.

Left-aligned image

The frame has a slightly blotchy appearance when viewed at high contrast. These blotches (assumed to be due to photo-response non-uniformity) vary on a scale of about 150 pixels or so, but never deviate from the mean value by more than 0.25%. (The Kodak specification suggests a nominal value of 1%). Uniform vertical banding, evident in the photon transfer analysis, is also visible. This has a peak-to-peak variation of typically 0.15%. This appears to be a fixed pattern, presumably associated with the manufacture of the chip, and should be removable by flat-fielding.

All of the rows of the flat field frame were added together and then divided by the total number of rows to obtain an average horizontal section. Part of this section is shown below, where the periodicity of the banding at 7 or 8 columns per band is clearly seen.

Left-aligned image