Development of Jatropha Curcas Color Grading System for Ripeness ...

European Journal of Scientific Research
ISSN 1450-216X Vol.30 No.4 (2009), pp.662-669
© EuroJournals Publishing, Inc. 2009
http://www.eurojournals.com/ejsr.htm


Development of Jatropha Curcas Color Grading System for
Ripeness Evaluation


Zulham Effendi
Mechanical and Material Engineering Department, Faculty of Engineering
and Built Environment, Universiti Kebangsaan Malaysia
43600 Bangi, Selangor, Malaysia

Rizauddin Ramli
Mechanical and Material Engineering Department, Faculty of Engineering
and Built Environment, Universiti Kebangsaan Malaysia
43600 Bangi, Selangor, Malaysia
E-mail: rizauddin@vlsi.eng.ukm.my

Jaharah Abdul Ghani
Mechanical and Material Engineering Department, Faculty of Engineering
and Built Environment, Universiti Kebangsaan Malaysia
43600 Bangi, Selangor, Malaysia

Zahira Yaakob
Chemical and Process Engineering Department, Faculty of Engineering and
Built EnvironmentUniversiti Kebangsaan Malaysia
43600 Bangi, Selangor, Malaysia


Abstract

This paper presents an automated grading system for Jatropha curcas by using a
color histogram. Today, Jatropha curcas have been widely planted for harvesting its nuts
in the biodiesel production. The sizable Jatropha curcas plantations are only established
when the world oil crisis demands extra capacities of biodiesel feedstock. As biodiesel
feedstock, the Jatropha curcas gives a prospective value because it is categorized as non-
edible oil that the availability will not be threaten by food purposes. Currently, human
experts perform the traditional identification of Jatropha curcas. Its quality depends on
type and size of defects as well as skin color and fruit size. In this paper, we present the
development of a Grading System of Jatropha (GSJ) by using color histogram method to
distinguish the level of ripeness of the fruits based on the color intensity. GSJ employs a
Mean Color Intensity(MCI) to analyze the red, green and blue (RGB) color of Jatropha
curcas. After few simulations, it can be ascertained that our proposed GSJ is useful for
grading the level of ripeness of Jatropha curcas.


Keywords: Jatropha curcas, Image Processing, Grading System, Color, Ripeness

---

Development of Jatropha Curcas Color Grading System for Ripeness Evaluation
663
1. Introduction
In spite of the world demand on the biodiesel stocks, Jatropha curcas which is a hardy shrub and
traditionally known in many subtropical and semi-arid regions for its medicinal properties has been
largely planted. It is widely used in the form of protective hedges around fields, to prevent animals
from grazing crops. As Jatropha curcas seeds and green leaves are poisonous, it works as a very
effective barrier. Long qualified as an interesting but "underutilized" crop, it is now being increasingly
used in reforestation programs in tropical countries because it thrives on poor soils and on land that is
suffering under erosion [1]. The good news for the world that is facing significant reduction in fossil
fuel availability as main source of energy because from Jatropha curcas, a biofuel can be extracted.
The use of fossil fuel also polluting the environment with carbon dioxide (CO2) which in turn causing
global warming. Therefore, seed oil from Jatropha curcas offers an excellent alternative for major
source of renewable energy in the drier rural areas of (sub)tropical Asia, Africa and America [2].
Traditional identification of Jatropha curcas is performed by human experts [3]. Color grading is an
important process for the Jatropha curcas . The color of Jatropha curcas is often used to determine
quality. Color grading applications are implemented by using color image processing [4]. Since food
products can be graded by their color, color grading for apples [5, 6, 7], strawberries [8], tomatoes [9],
and other fruits have been developed [10].
The shapes cannot estimate the quality of Jatropha curcas because the fruits may have a
different shape but the same level of quality [11]. In order to evaluate its quality, a grading system that
should be able to analyze the color of the Jatropha curcas and then obtain its quality based on density
of color by using red, green, and blue (RGB) model is needed. RGB color space is commonly used in
image processing because of its basic synthesis property and direct application in the image display.
Individual RGB channels are usually digitized into 8-bit values (0 to 255) and 24 bits are required to
represent the color value of each pixel. Although 24 bits allow the specification of more than 16
million unique colors, the range of distinguishable colors for any grading application is much smaller.
If the range of colors of interest is well defined for a given application, color grading can be
significantly simplified. RGB is the most commonly used color spaces for color grading because their
electronic images are often acquired in terms of RGB components. Color grading in the RGB color
space is complicated due to the difficulties of selection and adjustment of color preferences of 3D color
representations [11]. However, color is a good indicator to differentiate the ripeness of Jatropha curcas
and by using a digital image processing where the color data from a sample, the level of ripeness can
be retrieved without running a physical experiment on the sample [12]. In this paper, we present the
development of a Grading System of Jatropha (GSJ) by color histogram.


2. Color Histogram
Color histograms are easy to compute, and they are invariant to the rotation and translation of image
content. However, color histograms have several inherent problems for the task of image indexing and
retrieval. The first concern is their sensitivity to noisy interference such as lighting intensity changes
and quantization errors. The second problem is their high dimensionality on representation. Even with
coarse quantization over a chosen color space, color histogram feature spaces often occupy more than
one hundred dimensions (i.e., histogram bins) [13] which significantly increases the computation of
distance measurement on the retrieval stage. Color histograms are commonly used to represent images
because they provide a far more compact overview of the data in images than knowing the exact value
of every pixel, and they are invariant with translation and rotation about the viewing axis [14]. A color
histogram is derived by counting the 'color' of each pixel. Under the R, G and B color space we
decompose an image into these individual colors and build a histogram for each of them.
Some applications of color histogram are in photography, color histograms in either 2D or 3D
spaces are frequently used in digital cameras for estimating the scene illumination, as part of the
camera's automatic white balance algorithm. In remote sensing, color histograms are typical features
used for classifying different ground regions from aerial or satellite photographs. In the case of multi-

---

664
Zulham Effendi, Rizauddin Ramli, Jaharah Abdul Ghani and Zahira Yaakob
spectral images, the histograms may be four-dimensional, or more. Furthermore, color histograms can
be used in object recognition and image retrieval systems or databases in computer vision [15]. Also in
one large scale image database application, over 15000 images could be queried in under two seconds
by refining color histograms using a technique called "color coherence vector" [16]. Some approaches
exploit the color histogram derived together with a similarity measurement chosen to make color
histograms more robust to noisy interference. In order to identify objects based on their color
histograms, Swain and Ballard proposed a histogram intersection method, which is able to eliminate
the influence of color contributed from the background pixels during the matching process in most
cases [17]. Although their method is robust to object occlusion and image resolution, it is still sensitive
to illumination changes.
Funt and Finlayson [18] proposed a color constant color indexing method to extend Swain and
Ballard’s color indexing method by establishing the histogram of color ratios. Since the illumination
remains essentially constant locally, calculating the ratios of neighboring colors removes the
illumination variation component. Similar extension can be found in Drew et al.’s work [18].


3. Grading Methodology of Jatropha
This section describes the methodology for processing and analyzing Jatropha curcas images of GSJ.
The process involved in GSJ can be divided into three stages that is the image collecting stage, training
stage and testing stage. The schematic flow of the image processing is shown in Figure 1.

Figure 1: Flow diagram processing

Camera
Image processing
Data Analysis
Color Matching
Result


3.1. Images Collecting Stage
The first stage of GSJ is the images collecting stage where few samples of images are captured. In our
research, all of the Jatropha curcas samples were taken from Jatropha curcas plantation in Malaysia.
In general, the Jatropha curcas are classified into three different grades that are raw, ripe and overripe
[19].
For sampling purpose, we used 5 Jatropha curcas of each grade in training stage and 6 samples
for the testing stage. From these samples, the color images of Jatropha curcas are collected by using a
digital camera and saved in JPEG format. These images are stored in the hard disk of memory card and
transferred to hard disk of computer. The sample images of each grade are shown in Figure 2 (a), (b)
and (c).

---

Development of Jatropha Curcas Color Grading System for Ripeness Evaluation
665
Figure 2: Sample data for each grade (a) raw (b) ripe (c) over ripe


(a)
(b) (c)



3. 2. Training Stage
The second stage of GSJ is the training stage. In this stage, the analysis of the RGB data of Jatropha
curcas images is carried out by using software called ImageJ. In the software, the mean color intensity
is used to distinguish between the ripeness of the fruits by using the following equations.
x =
P /
P (1)
r
∑n ri ∑n i
i =1
i =1
x =
P /
P (2)
g
∑n gi ∑n i
i=1
i=1
x =
P /
P (3)
b
∑n bi ∑n i
i=1
i=1
where,
x , x , x is the mean of red, green and blue pixel respectively , P , P , P is the number of pixel
r
g
b
r
g
b
for red, green and blue respectively and Pi is the total number of pixel in the image.
In this training stage, the calculation of color intensity is performed on each image of the
Jatropha curcas. The means for entire fruit are obtained by calculating the means for all images of
grades. This process is repeated with all the samples that are used as training samples for the program.
The color of Jatropha curcas is classified to raw, ripe and over ripe. Figure 3, 4 and 5 show the
samples in each class, respectively. In order to determine the classification of the Jatropha curcas, we
adopt the calculation of the range value, i.e., the minimum and maximum mean of RGB density for
each category. This range value is used as the reference and standard value of grading for our test
program.

Figure 3: Raw images


(a)
(b)


Figure 4: Ripe images


(a) (b)

---

666
Zulham Effendi, Rizauddin Ramli, Jaharah Abdul Ghani and Zahira Yaakob
Figure 5: Over ripe images



(a) (b)


3 3. Testing Stage
The final stage of GSJ is the testing stage. In this stage, the program calculates the mean color intensity
for RGB, in which is related to the mean color intensity for RGB of Figure 3(a), 3(b), 4(a), 4(b), 5(a)
and 5(b) individually. After that, the mean color intensity of RGB for all these 6 figures are calculated
and the mean of all figures are computed. Next, after the calculation of the mean color intensity of
RGB, the program runs a test for the grading the Jatropha curcas which is based on the standard mean
of the grading according to the training program.


4. Results of Grading
The computer program calculated the mean color intensity to differentiate between the different color
or grading of the Jatropha curcas. The mean RGB for each category of the Jatropha curcas sample is
shown in Table 1, 2 and 3 respectively. After obtaining the RGB intensity of the samples for each
category, we compared them with each other in order to calculate the range of RGB intensity of the
Jatropha curcas for each category as shown in the Table 4. These ranges are placed in the testing
program as reference of the grading for testing of Jatropha curcas..

Table 1:
Mean RGB intensity of the sample for raw category

Mean
Sample
Red
Green
Blue
1 215.91
218.93
206.29
2 222.59
225.18
216.56
3 227.88
230.81
214.38
4 228.37
232.99
218.75
5 220.82
225.59
211.78

Table 2:
Mean RGB intensity of the sample for ripe category

Mean
Sample
Red
Green
Blue
1 235.09
229.65
211.5
2 242.52
239.05
229.25
3 237.34
233.8
223.16
4 223.54
219.11
206.2
5 220.82
225.59
211.78

---

Development of Jatropha Curcas Color Grading System for Ripeness Evaluation
667
Table 3:
Mean RGB intensity of the sample for over ripe category

Mean
Sample
Red
Green
Blue
1 224.25
222.79
222.36
2 200.31
198.49
197.55
3 216.37
214.77
213.87
4 205.78
204.63
203.93
5 217.79
216.83
216.17

Table 4:
The range of RGB intensity of the sample of category

RGB Intensity Range
Category
Red
Green
Blue
min
max
min
max
min
max
Raw
215.91 228.37 218.93 232.99 206.29 218.75
Ripe
220.82 242.52 219.11 239.05 206.2 229.25
Over
ripe 200.31 224.25 198.49 222.79 197.55 222.36

The mean of RGB for each picture that has already been obtained through the procedures. The
mean of RGB for the entire fruits are obtained by calculating the mean value for the two images that
are shown in the Table 5.
Finally, we tested the Jatropha curcas depends on the standard mean of the grading. From our
present experience in this research, the accuracy of the result obtained depends on the accuracy of the
data source and the number of training samples. The grading system is able to detect and distinguish
different categories of classification of the Jatropha curcas. Figure 6(a), (b) and (c) show examples of
the detection results.

Table 5:
The mean of RGB intensity of the tested fruit

Mean
Images
Grade
Red
Green
Blue
1 223.24
226.87
214.82
Ripe
2 215.54
218.51
206.42
Over
Ripe
3 248.65
246.02
235.25
-
4 246.57
243.65
227.8
-
5 215.46
214.02
213.51
Raw
6 228.56
227.61
227.25
Ripe


5. Summary and Concluding Remarks
This research is conducted in order to differentiate between the colors properties of Jatropha curcas.
The RGB data images of the Jatropha curcas are analyzed by a computer program that is written using
ImageJ software. Three types of classification is used to grade the ripeness of Jatropha curcas that is
raw, ripe and over ripe. As the results of the training stage, the range of RGB intensity of the Jatropha
curcas can be used as the reference of the grading in the testing stage. In the future, the grading system
will be developed as an integrated program as a user-friendly application that allows the farmers to
determine the grades of their Jatropha curcas.

---

668
Zulham Effendi, Rizauddin Ramli, Jaharah Abdul Ghani and Zahira Yaakob
Figure 6: Example of Jatropha curcas image testing (a) Image (b) result value (c) result histogram


(a)


(b) (c)



References
[1]
Gübitz, G.M., Mittelbach, M. and Trabi, M., 1999, Exploitation of the tropical oil seed plant
Jatropha curcas L., Bioresource Technology, Vol. 67, pp. 73–82.
[2]
Fangrui M. and Milford A.H., 1999. Biodiesel production: a review. Bioresource Technology,
Vol. 70, pp.1-15.
[3]
R. E. Gonzalez and R. E. Woods, 1993, Digital Image Processing, Addison-Wesley.
[4]
J. Blasco, N. Aleixos and E. Molt´o, 2003, Machine vision system for automatic quality grading
of fruit, Biosystems Engineering, Vol. 85(4), pp. 415–423.
[5]
D. Unay and B. Gosselin, 2007, Stem and calyx recognition on 'Jonagold' apples by pattern
recognition, Journal of Food Engineering, Vol. 78(2), pp.597-605.
[6]
D. Unay and B. Gosselin, 2004, A quality sorting method for `jonagold` apples, Proceedings of
International Conference of Agricultural Engineering Leuven 2004-Engineering the Future,
Leuven, Belgium.
[7]
K. Nakano, 1997, Application of neural networks to the color grading of apples, Computers and
Electronics in Agriculture, Vol 18, pp.105–116.
[8]
Nagata M;, Kinoshita O;, Asano K., et al, 1997, Studies on automatic sorting system for
strawberry (Part 3)”. Journal of Japanese Society of Agricultural Machinery, Vol.59(1), pp.43-
48.
[9]
K.H. Choi, G.H. Lee, Y.J. Han and J.M. Bunn, 1995, Tomato maturity evaluation using color
image analysis”, Trans. of ASAE, Vol. 38(1), pp. 171-176.
[10]
J.B. Njoroge, K. Ninomiya, N. Kondo and H. Toita, 2002, Automated fruit grading system
using image processing”, Proceedings of SICE 2002, Osaka, Japan. pp. 1346-1351.
[11]
Lee, D.J., Chang Yuchou, et al, 2008, Color Quantization and Image Analysis for Automated
Fruit Quality Evaluation, Proceedings of 4th IEEE Conference on Automation Science and
Engineering, Washington DC, USA, August 23-26.
[12]
Jain, A.K., Vailaya, 1996, A Image retrieval using color and shap, Pattern Recognition, Vol.
29, pp. 1233-1244.
[13]
J. R. Smith and S.-F. Chang, 1995, Single color extraction and image query,” Proc. ICIP, Vol.
3, pp. 528–531.
[14]
W. Lidan, 2006, Clustering www image search results using color histogram and textual
information, CS838 Term Project Report, May 2006

---

Development of Jatropha Curcas Color Grading System for Ripeness Evaluation
669
[15]
M. Stricker, and M. Swain, 1994, The Capacity of Color Histogram Indexing, Proceedings
1994 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, pp.
704-708
[16]
Pass G. and Zabih R., 1996, Histogram refinement for content-based image retrieval,
Proceedings 3rd IEEE Workshop on Applications of Computer Vision, 2-4 Dec 1996, pp. 96-
102.
[17]
M. J. Swain and D. H. Ballard, 1991, Color indexing, International Journal of Computer
Vision, Vol. 7(1), pp. 11-32. .
[18]
M. S. Drew, J. Wei, and Z.-N. Li, 1998, Ilumination-invariant color object recognition via
compressed chromaticity histograms of color-channelnormalized images, Proc. 6th
International Conference of Computer Vision, pp. 533–540.
[19]
Sumanto, 2006, Pengaruh ketuaan buah jarak pagar terhadap kandungan minyak, Jurnal Infotek
Jarak Pagar, Vol.1 (3), pp.3 (in Indonesian).

---