MODULE VI

OPTICAL PROPERTIES
OF AB MATERIALS

Color Theory and Models

Module
VI
 136

MODULE VI

OPTICAL PROPERTIES OF AB
MATERIALS

 INTRODUCTION

This module deals with the optical properties specifically the

color theory and models for AB materials. As an ABE student, it is
expected that you will see and learn how important or significant it is
to know or study the color theory and models for AB materials.

LEARNING OBJECTIVES

After studying the module, you should be able to:

1. identify the optical properties specifically the color theory

and models of AB materials; and
2. determine and explain measurement methods of
measuring or identifying the optical properties of AB
materials.

 DIRECTIONS/ MODULE ORGANIZER

This module composed with one lesson. Read and analyze the
lesson carefully then answer the activities/ exercises. Work on these
activities carefully and submit your output to your subject instructor on
or before the set deadlines.

Module
VI
 137

 COLOR THEORY AND

MODELS
Color and appearance of AB products such as fruits and
vegetables are perhaps the first attributes that attract us. Since the
appearance of the product generally determines whether a product is
accepted or rejected, measuring the color characteristics becomes an
important task. To carry out the analysis of this key attribute for
agriculture, it is recommended to use an artificial vision system to
capture the images of the samples and then to process them by
applying colorimetric routines to extract color parameters in an
efficient and nondestructive manner, which makes it a suitable tool for
a wide range of applications.

The color assessment of fruits and vegetables in the food

industry and agriculture using machine vision, image processing and
spectroscopy has become a trend in the recent years. The color
features are one of the key parameters to define the quality of an
agricultural product. The color is probably the first factor that
consumers use to determine the appearance of a product; appearance
is a subjective factor that leads the consumer to accept or reject a
food product. This significantly affects the sales and profits of the
industry. Therefore, a considerable effort has been made in the area of
automation to improve the quality of agricultural products in the food
industry in order to decrease losses.

Building machines with the ability to see color as it does the

human being has been a complex task for the scientific community and
industry in recent years. Among the many challenges to be addressed,
we can include appropriate image acquisition systems, lighting
problems, color space definitions, mathematical issues, and the
development of specific algorithms and synchronization tasks.
However, improvements in semiconductors, electronics, and software
eventually brought the opportunity to implement image processing and
colorimetric projects for various applications.

A machine vision system for horticulture products requires the

ability to capture process and analyze color images, where algorithms
are suitable to detect, extract, and quantify the attribute of color as
much as a customer does. Furthermore, other parameters (size,
texture, external blemishes, and diseases) are important to determine
the appearance, and hence the quality of an AB product. This variety of
applications is possible due to the interaction of light in the range of

Module
VI
 138
visible spectrum (400–700 nm) with the matter that the light can be
reflected, transmitted, or absorbed by an object. The light wavelengths
received by our eyes are then interpreted by our brain as color.

Perception of color, in humans, is a psychophysical phenomenon

that involves three elements: the illuminant, an object, and the
observer. The

Module
VI
 139

illuminant has the function to irradiate the object with light in the
visible spectrum. The object absorbs, transmits, and reflects the light
received from the illuminant. The observer perceives the reflected light
from the object in the retina and responds to that stimulus generating
an electric signal in the optic nerve toward the brain. As this
phenomenon takes place in the brain, the process is an important
challenge, because machine vision systems need to emulate these
three elements, in a proper way, without including the real observer,
the eye, and the natural illuminant, the sun.

A machine vision system acquires, processes, and analyzes

images, and the proper operation of the system requires adequate
capture conditions for the specific application. Basic machine vision
components are: (a) image sensor (charge-couple device (CCD) or
complementary metal oxide semiconductor (CMOS)), (b) illuminant (D-
65, fluorescent lamps, among others), (c) the background (with high
contrast for the object of interest), (d) trigger device (used to start
image acquisition), and (e) a frame grabber (to capture the actual
image). The capture procedure can be divided into the following steps:
the illuminator irradiates the object, the image sensor receives the
light reflected from the object and the background, and when the
trigger is activated, the system extracts the color characteristics and
converts them to electrical signals and then the grabber frame stores
the image taken. Once the image is on the computer or other
processing device, image processing algorithms are used to analyze
the data. The basic steps can be listed as: (a) segmentation of images
(the background is separated from the region of interest), (b)
extraction of characteristics (select pixels of interest); for this stage,
several functions and algorithms have been developed in order to get
an adequate image processing of the scene. Next, we analyze the data
generated from the previous process, the system extracts the color
attribute of the objects in the image using colorimetric techniques, as
well as others as morphological and texture parameters, and finally the
analysis results are presented.

Colorimetry is the scientific area that measures, quantifies, and

represents color. It is very useful in different areas because it provides
the ability to turn color into an objective factor rather than a subjective
one. Complementing this scientific area with the technology of
computer vision systems, it is possible to see the color characteristics
in a digital and standard way, in addition to performing the color
evaluation with a noninvasive and noncontact procedure, which makes
it a suitable technology for application in agriculture and the food
industry for quality assessment. Due to their versatility, many other
production systems can benefit from the features offered by artificial
vision systems, image processing, and colorimetry.

Module
VI
 140

Method of Approach to a Color Problem

There are two distinct methods for measuring color. The first
method is that the stimulus of any color may be completely specified in
terms of intensity at each wavelength or frequency or partially
specified in terms of secondary standards and the other method is that
colors may be matched by the use of secondary standards and
specified in terms of color, not in terms of the stimulus.
According to the report of the colorimetry committee of the
Optical Society of America, color is the general name for all sensations
arising from the activity of the retina of the eye and its attached
nervous mechanisms, this activity being in nearly every case in the
normal individual, a specific response to radiant energy of certain
wavelengths and intensities. Further report says that color is
fundamentally a psychological category, which means that we cannot
see wavelengths but that we can see color. To measure color we must
deal with it psychologically – in terms of what we see, not in terms of
the wavelength stimulus.

Color may be completely specified in terms of three attributes:

Hue, brilliance, and chroma. Hue is the attribute which permits colors
to be classed as reddish, yellowish, greenish, or bluish; brilliance is the
attribute by which a light color is distinguished from a dark color or
brilliance is that attribute of any color in respect of which it may be
classed as equivalent to some member of a series of grays ranging
between black and white; and chroma is the attribute by which a
strong color may be distinguished from a weak color or chroma is that
attribute of all colors possessing a hue, which determines their degree
of difference from a gray of the same brilliance. Each of these
attributes may be measured on a scale of psychologically equal steps.

Regarding the measurement of color, the three attributes of

color can be treated as quantities and specified numerically, if all
discriminable colors are conceived to be arranged into a system such
as that neighboring members differ from one another in each of the
three attributes by just noticeable degrees. Such a system is
necessarily three-dimensional, and three ordinal values representing
the positions of a given color in the several dimensions are needed to
define the color.

With such a definition of color and its attributes, and with a

method of measurement so obvious in its simplicity, it would seem that
any colorimetric method of measuring color to be employed in
standardizing and studying a series of colors should be based on
measurements of hue, brilliance, and chroma.

Module
VI
 141
Both spectrophotometric and colorimetric measurements may be
made in different ways. Circumstances may make one type of color
measurement preferable to any other. Therefore before a method is
chosen the problem

Module
VI
 142

should be carefully studied. If the problem concerns the specification of

colors already produced, so that one can record what the look like, or if
it concerns colors which are not to be combined to produce new colors,
then it is probable that descriptive color measurements are desirable—
that is, measurements in terms of hue, brilliance, and chroma. But if
the problem concerns the stimulus of the color, how it is produced,
which wave lengths are most highly reflected, and how the colors will
combine to produce new colors, then stimulus measurements are
desirable—that is, measurements in terms of intensity at each wave
length or group of wavelengths.

Color Problems of ABE Products

The element of color is an integral grading factor in numerous

standards established for AB products. Cotton, for instance, is sold
according to grade and staple. The grade factor consists of three
variables: Color, leaf and trash, and preparation or ginning. Hay is
graded on color, foreign material, and condition. Color is an important
element in grading fruits and vegetables; it is a part of the
specifications for cotton-linters standards; it plays a part in grading
rice, honey, meat, grains, breads, mayonnaise, and innumerable other
agricultural products, or products made from those of agriculture,
often with direct correlation in protein content, diastatic activity, or
money value.

Image Processing

The techniques of image processing are related to algorithms

that manipulate the numerical representation of the images to obtain
useful information. In particular, segmentation subdivides the digital
image into multiple regions or objects that have common
characteristics. This action involves several steps that must be taken,
before the images can provide valuable data. First, region of interest
(ROI) must be identified, i.e., regions in the image that have pixels that
matter to the application and which must be separated from the
background. In order to develop algorithms that perform this work, it is
necessary to define whether the image is black and white or colored,
and then a criterion must be established for the segmentation, which
must focus on morphological, texture, or colorimetric parameters,
before establishing any flow chart as solution. It should be mentioned
that algorithms may be suitable for many applications, but none of
them is generally applicable to all images, and therefore, suitable
algorithms for each particular application should be used.

Image Acquisition System and Color Image Representation

Module
VI
 143

It is essential to use an image acquisition system that suits the

application properly. It must be defined what type of objects or
products will be placed in the system, as well as set the scene
location and choose the

Module
VI
 144

appropriate settings to obtain acceptable images. The issue of lighting

(natural or artificial) depends on the scenes required by the
application. Once the capture conditions are adequate, a vision sensor
(CCD or CMOS) must be selected, which will suit the entire system for
proper performance. A digital camera receives light variations
corresponding to images onto a CCD device. The CCD contains
capacitors that are stimulated by visible radiation and three filters
adjusted for three basic colors: red (R), green (G), and blue (B).
Theoretically, every color can be reproduced by the combination of
three primary colors.

A color image are represented as a M × N × 3 (components)

array of color pixels, where each color pixel is a triplet corresponding
to the RGB components of and image at specific spatial location, as
shown in Figure 1. By convention, the three images formed and RGB
color image are referred to as the R, G, B component images. The data
class of the component from the image determines their ranges of
values, for example [0–255] or [0–65,535] for RGB images of class
unit8 or unit16, respectively. The RGB color space is an additive color
model that uses transmitted light to display colors. It is used for
television and other devices screens; so this model is device
dependent (Its appearance depends on the display.)

Figure 1. Color representation of digital image and their RGB

components

Color Image Segmentation

Algorithms for color image segmentation based on the technique

of threshold segmentation should be executed in triplicate, due to the
structure of the color image. A typical segmentation process fills in
with ones and zeros of the image matrix locations, corresponding to
the selected regions in each of the color channels of an image, as
shown in Figure 2.
Module
VI
 145

Figure 2. Segmented image: a) background filled with zeros and

b) background filled with ones.

Color Calculations

The Commission Internationale de l’Eclairage (CIE) determines

regulations, standards, and recommendations for color measurements.
The CIELAB color space is an international standard developed by the
CIE in 1976. Within CIELAB, a psychometric index of lightness (L*) and
two color coordinates (a* and b*) are defined. L* is a qualitative
attribute of relative luminosity, which is the property according to
which each color can be considered as equivalent to a member of the
gray scale, ranging between black (L* = 0) and white (L* = 100).
Negative values from a* correspond to greenish and the positives to
the reddish ones, whereas the yellowish colors takes negatives values
from b* and the positive for bluish ones.

It is well known in the food industry that the CIELAB color space
is used to analyze color changes in a qualitative way. Color and
appearance are closely related to the sensory properties and chemical
composition of food. Color is usually measured by tri-stimulus
colorimetry. The color stimulus is composed of three different
parameters, giving the color a three-dimensional nature.

The color attributes are described in CIELAB as:

 Lightness: This feature indicates if a color is lighter or darker. It is
a relative measure of the reflected light against the absorbed.
Value 0 corresponds to black and value 100 is assigned to white.
 Chroma: It determines for each hue, the color difference taking
as reference the gray level with same lightness. It can take

Module
VI
 146
positive values from zero.

Module
VI
 147

 Hue: It is the main attribute. It is a qualitative property, which

allows classifying colors as red, yellow, etc. It is related to
differences in absorbance of radiant energy at different
wavelengths. Hue is specified as an angle.

These attributes are often expressed as L*, Cab*, and hab°,

respectively; according to the CIELAB color space, it can be
represented as Cartesian coordinates of polar coordinates C ab* and
hab°. It can be used on a variety of instruments, such as colorimeters,
spectrophotometers, and spectroradiometers. However, these
instruments require homogenous samples to achieve a uniform color,
which becomes a tedious and complicated task to measure the color of
heterogeneous or small objects, such as grape berries and grape
seeds. In these cases, the use of digital images for the extraction of
color characteristics is advantageous. Digital image analysis appears
as a suitable complement since it is possible to extract not only color
characteristics but also other characteristics such as shape, texture,
and homogeneity.
On the other hand, due to the nature of the CIELAB color space,
the calculation of the Euclidean distance can be applied between
neighboring samples, in order to obtain a relation between a
quantitative and qualitative value on the variation of the color
appearance of an object.

Application of Image Processing and Color Analysis to AB

Product (Sample Case Study: Habanero Chili Pepper as Sample
Product)

Since consumers buy with their eyes, color is considered one of

the most important quality parameters of food products. Normally, this
is determined by human inspection, or measured using a colorimeter
or a spectrophotometer. The first process is subjective and susceptible
to fatigue. The second is limited to measure just a small area of the
food product, making it difficult to obtain a clear view of the color of
the complete sample. In order to overcome these limitations, a system
of artificial vision, image processing, and color analysis has been
applied to measure the postharvest color of the Habanero chili
(Capsicum chinense Jacq.) at different stages of ripening.

The methodology used is described as follows: (1) specimens of

Habanero chili here collected and arranged in three groups depending
on their maturity stage by visual inspection; (2) The samples are

Module
VI
 148
moved to the laboratory to be placed in the machine vision system and
the acquisition process is setup; (3) The acquisition process starts with
a 24 sampling rates and finishes after 15 days, capturing a single
image from each specimen; (4) Once the images were acquired, a
database with a total of 900 images was generated; (5) The
algorithms of image processing and color analysis were

Module
VI
 149

applied to this dataset; and (6) Finally, the algorithms generate results
with color segmentation, colorimetric measurements in CIELAB color
space, color analysis, and statistical analysis of the dataset.

For this study samples of Habanero chili were harvested from an

aquaponic greenhouse culture. The selection of the samples was
carried out by a specialized technician. The expert harvested a
representative group of Habanero chili, which showed different stages
of maturity. A color categorization was performed by visual inspection.
The samples were separated into three groups of colors (green,
yellowish and orange), with 20 specimens per group.

Set-up of Machine Vision System

The artificial vision system used a color CCD camera as image

capture device. The lighting system contained fluorescent lamps
mounted on top to avoid shadows. In order to control the camera and
to download the images, a standard PC with MATLAB was used. The
system design allowed the camera to move the samples above to keep
them stable. Because of the protrusions and cavities presented by the
Habanero chili, it is difficult to keep the surface of interest in exactly
the same position along the experiment. As shown in Figure 3, the
samples were placed in the confined space of the machine vision
system to acquire the corresponding images.

Image Processing Algorithms

To accomplish the image processing tasks, the flow chart was

followed as showed in Figure 4. The algorithms are executed using
suitable functions in order to get reliable information from digital
images. At the beginning, the image from sample is captured using the
image acquisition system. Then, image processing routines carry out
the required operations to separate the ROI, as well as to convert the
image from RGB color space to CIELAB. The calculations to obtain the
CIELAB components (L*, a*, b*, Chroma, Hue angle) using tri-stimulus
colorimetry are performed to generate the corresponding array.
Finally, the color analysis from the Habanero chili in postharvest
conditions can be accomplished.

Module
VI
 150

Figure 3. Proposed Set-up for the Case Study.

Figure 4. Flow Chart used for the Image Processing

Algorithm.

The original and processed images, by the algorithms from the

dataset, are shown in Figure 5. One image, aleatory selected, from
each categorized group (green, yellowish, and orange) is presented. At
the top, you can identify the categorization of groups, and below is the
corresponding label of the sample. Then, the original images, just as
the vision system acquired them, are presented. In the middle, the
label that indicates the exact day of the acquisition is located. At the
bottom, the processed images are shown, where you can clearly
observe the segmentation process and how the algorithms filled
Module
VI
 151

the background with zeros, and only the information from the region
of interest is processed (ROI).

Figure 5. Original and Processed Images of Habanero

Chili
Color and Statistical Analysis
Habanero chili is a climacteric fruit, which means that once it is
cut, it begins to ripen. Depending on the variety of Capsicum chinense
Jacq., the color of the Habanero chili changes during maturation. In
general, changes occur from the dark green, in its initial stages, and
then go through the yellowish-green until reaching an orange color, in
the final stage of maturity.
Color evolution can be represented as hue angles (h ab°) in a polar
graph. Typically, Habanero chili initiates maturity with green stage
(under the threshold of hab = 120° for green) and then it moves
through the yellowish stage (between the threshold h ab = 120 and 60°
for yellowish) to achieve orange colors (crossing the threshold of h ab =
60° for orange) in the final stage of maturity, as time passes. The color
information from the dataset showed this behavior, and the statistical
analysis presented in Figure 6 demonstrates that image processing
and colorimetry are capable of extracting reliable values from acquired
images and detect color changes from these agriculture products.
Therefore, this methodology described, as noncontact technique, can
be considered a suitable option to analyze the color of Habanero chili.
In Figure 6, chart (a) shows the variation from the green group.
In average values of hue angle, a high color change is presented, due
to the gradual transition from green colors to oranges passing through
the yellowish ones during their maturity process. In chart (b), a
descendant gradual color change can be appreciated, even when

Module
VI
 152
crossing the threshold of h ab = 60°. Instead, the third group (c),
corresponding to the orange ones, remains after the threshold of h ab =
60° with a slow progressive color changes.

Module
VI
 153

A one-way ANOVA was conducted to evaluate the relationship

between color changes in 20 Habanero chili and 15 days of sampling.
Table 1 displays the summary for the one-way ANOVA for each group
of samples. With the null hypothesis (H 0), it was showed that all the
color values are equal during 15 days. In the analysis, it was shown
that in green, yellowish, and the orange groups, the ANOVA was
significant: F(14, 285) = 8.17, F(0.05) = 1.692, F(14, 285) = 100.21,
F(0.05) = 1.692, and F(14, 285) =8.17, F(0.05) = 1.692,
respectively. The ANOVA results allowed to reject the null hypothesis
and supported the conclusion that there is a statistically significant
color change during days for the green, yellowish, and orange groups.

Figure 6. Box and whiskers chart from each group: a) green, b) yellowish,
and
c) orange.

Module
VI
 154

Table 1. One-way ANOVA table for each group

Conclusions
Artificial vision systems combined with image processing and
color analysis are a reliable and affordable option when specific
applications require the use of non-invasive and noncontact
techniques. Similar characteristics of the samples are extracted from
their images and grouped for further analysis using image processing
techniques, which helps to obtain consistent and reliable separations of
elements. The CIELAB color space provides the parameters needed to
analyze and calculate important characteristics of a color image. Color
differences can be detected more directly using the CIELAB color
space, and it is important to mention that a color difference magnitude
can be imperceptible to the naked eye, but it is a basic operation for
the vision systems. However, the context in the color analysis should
be considered as an important factor, for the proper interpretation of
the data generated from the previous process. For example, the color
attribute of the Habanero chili is a fundamental parameter for the
appearance of the genus capsicum, which can be evaluated by image
processing and colorimetry to detect color changes with adequate and
reliable results in postharvest analysis. Trend in applications of color
analysis and image processing for agriculture will continue to increase
in the near future, due to the great variety of colors and shapes of the
products, in particular, the interest to obtain the best quality.

Module
VI