A flat-field or uniform photometric response is achieved by calculating reflectance factors
(RF), which are the ratios of the reflected radiation from a sample (RS) to those of a known
reference material (RR). An RF value for an individual pixel (i) at a given wavelength (A) can
be calculated using following equation:
ary, a RSA RDIA Rex
VM RRiA — RDA >
where, RS =sample image
RD =dark current image
RR =reference image
RC =correction factor for the reference panel
This considers that the signal from each pixel will consist of a gain (a signal given by the
detector which varies as a function of the amount of radiation, RS or RR) and a dark current
(a signal given out by the detector when there is no incident radiation, RD). To eliminate the
impact of the dark current signal, the dark current image is subtracted from the gain. The
ratio of the signals for the sample and reference image is then multiplied by a correction
factor RC. An RC of 1.0 was used in our applications so as not to change the overall
brightness of the image.
Flat-field corrections often make use of an image captured with the sensor in the dark by
leaving the lens cap on the camera (a “lens-cap black”) to access the impact of the dark
current signal. Experiments allowing for a lens-cap black were conducted during our
investigations, but the improvement was very marginal with the cameras tested. As a result,
a dark current image subtraction was not included in the approach adopted. However, to
check if this assumption is pertinent to the camera employed, simply take an image with the
lens-cap on and look at the values for the image. If the image is below the noise level then
the dark current signal will not have a significant effect on the flat-field correction.
Sensor size and impact on flat-field corrections
Flat-field correction is particularly important in images taken with cameras having full-frame
sensors, where effects such as lens vignetting (a reduction in the brightness or saturation of
an image at the periphery compared to the image centre) are more prevalent. In general, the
cameras tested did not have full-frame sensors and only use the centre 50% of the lens. In
such cases, although still recommended, flat-field correction may be omitted in
circumstances where taking an image of a uniform reflective card may not be possible.
Provision in the software design will be made to accommodate such situations.
If using cameras with full-frame sensors this correction may be particularly important.
Preliminary tests may need to be done on cameras with such sensors in order to quantify the
impact of such effects.
Version No. 1.0 37 Date : 14/10/2013