Important Parameters for Image Color Analysis: An Overview - Visual Computing: Scientific Visualization and Imaging Systems - page 87

from the SPD of the source, and is an average calculated from the difference in
color of a prespecified set of samples under both a reference illuminant and a test
illuminant [ 7 ]. The best performance of a light source, when compared to a given
reference, generates a CRI of 100. A CRI greater than 80 is generally acceptable
depending on the requirement of the application that a lamp must meet [ 8 ]
(Table 4.1 ).
The types of illumination most commonly used are fluorescent lamps (65 %),
incandescent bulbs (10 %), natural lighting (19 %), and LED (2 %) [ 1 ].
Undoubtedly natural lighting would be the best option; however, its characteristics
widely changes depending on the sun angle, time of day, weather conditions, etc.
Incandescent lamps have a high CRI value, but have high energy consumption and
its usage has been avoided. The technology of fluorescent lamps has replaced
incandescent bulbs, but their characteristics are varied depending on the compo-
sition (amount or type of phosphorus) and it also contains mercury. Even so, its
consumption has greatly increased, especially with the use of compact fluorescent
lamps, due to their low power consumption and reasonable lifetime.
So, the lighting used to characterize a material must be selected regarding the
properties of the source (lamp) and its application in particular cases requires
experimental testing [ 9 ].
Mohan et al. [ 11 ] determined in a comparative study that the reflectance of
grains in the near infrared region allowed for a better classification accuracy than
the reflectance in the visible region, thus enabling the choice of the best source to
be used in the classification of grains. Manickavasagan et al. [ 12 ] studied the
influence of three types of lighting in the classification of grains: incandescent with
2,870 K, fluorescent ring with 3,000 K and fluorescent tubular with 4,100 K. They
both concluded that fluorescent lamps of the T8-type, normally used in classrooms,
have better accuracy for identifying defects in grains compared to halogen bulbs,
because these have a greater emission of radiation in the infrared region, thus
confirming the study by Mohan et al. [ 11 ].
The background color has a direct influence on the analysis of color, being
another important parameter for image acquisition. Research demonstrates that the
background where the material is analyzed can influence the perception of the
color of the material [ 13 - 15 ]. According to Brown and MacLeod [ 13 ], the per-
ceived color of a scene relies on the ratio between the light signals from that point
and light signals from surrounding areas of this scene.
Because the appearance of objects influences directly the consumer's decision
about the quality of the product, Dobrza ´ ski Rybczynski [ 16 ] studied the influence
of the packaging of fruits and vegetables in color perception. In this study, oranges,
carrots, beetroots, and parsleys were packed with nets of different colors in order to
analyze their influence on color and consumer preference. It was concluded that
the red colored net influences the color of the orange, appearing to be a more
mature fruit, influencing the final consumer.
Black or white backgrounds are the most used [ 17 ] because, for the purpose of
the analysis, they facilitate the segmentation stage. Blasco et al. [ 18 ] presented a