The colour of rice depends on many variables under which surface roughness and
translucency. To analyse the colour of a bulk sample, the rice has to be made opaque by
imaging a sufficient thick layer. For the analysis of the colour distribution of rice (within and
between rice kernels), an image has to be made of a single layer of rice kernels. The colour of
a single layer of transparent rice kernels is affected by the background used to cover these
kernels. FBS measures colour by using red, green and blue (RGB) filters. Each pixel in a FBS
image can display 256 levels (8 bit) of intensity (brightness value) for each of the three
colours, resulting in 16 million colours. The colour represented by RGB depends on the
scanner used. There are a number of colour scales (colour spaces or models) in addition to the
RGB colour scale. For rice commonly the L*a*b* colour scale is used. This scale is an
international standard for colour measurement developed by the Commission Internationale
de l’Eclairage (CIE) and is device independent. The CIELAB scale is based on a standard
source of illumination and a standard observer. It is a non-linear system as is the human
visual system. The three-dimensional L*a*b* colour scale consists of a luminance or intensity
component L* ranging from 0 (black) to 100 (white), along with two chromatic components
a* (degree of redness or greenness) and b* (degree of yellowness or blueness) ranging both
from –60 to +60 (see Figure 2). The a* and b* components define the colour and saturation.
The suffix * denotes that the CIELAB colour scale is used and not the Hunter Lab colour
scale [25]. The b values of the latter scale are very different from the b* values of the
CIELAB colour scale. The L*a*b* values can be calculated from RGB values after
calibration with standard samples measured by colorimetric or spectrophotometric methods
[30]. The RGB values measured by FBS are first transformed to XYZ tristimulus values