PERPUSTAKAAN UMP
CLASSIFICATION OF Kf 1 I IH 111111111
iE PROCESSING
0000103183
REWATHI D/O KARUDAN
A thesis submitted in partially fulfillment of the requirements for the award of degree of Bachelor of Science Computer (Graphic & Multimedia Technology) (Hons)
Faculty of Computer System & Software Engineering University Malaysia Pahang (UMP)
DECEMBER 2014
vi
ABSTRACT
Clarification of Key Lime base on color Image Processing is newly proposed system in the
Malaysian agriculture field. Therefore, this system carried out to develop a prototype judging the
Key Lime grade, maturity and to estimate the storage lifespan of Key Lime by color. The main
issue that inspired to develop this system is the manual practice of Key Lime color grading
consuming man power and time.At the same time the judgment of Key Lime color grading by
human eyes often leads to error due to visual stress, and tiredness and is therefore not accurate.
Other than that the human vision has limited ability in differentiating similar colors such as blue
and violet. The issue is a human perception towards colors is subjective and varies among
different people. Fifty five sample data of Key Limes were set in experiment to estimate the
expiry date of key lime. During the system execution, image acquisition phase the image of the
Key Lime should be in RGB color image. The grading systems computerized and web camera
are used to capture the Key Lime image. The background (noise) of the image is removed by
median filtering. A percentage of mean of green value was extract from RGB image for color
information and sum of threshold image was collected. Then values are then being used as
information for determining the grade and estimate expiry date of Key Lime. As conclusion the
system was tested using 55 sample Key Limes. The final result of testing showed that 90% pass
for grade A, 25 % pass for grade B, 71.43 % pass for grade C, 93.75% pass for grade D and
overall passing percentage is 80.39%.
vii
ABSTRAK
Pengkelasan limau nipis daripada wama boleh memberikan manfaat secara tidak langsung
kepada sektor pertanian di Malaysia. Oleh yang demikian, sistem mi telah dibangunkan satu
prototaip bagi menilai gred untuk menganggarkan tarikh luput limau nipis berdasarkan warna.
Faktor yang menyebabkan system mi dibangunkan ialah pengelasan secara manual mangambil
masa yang agak lama dan tenaga manusia.Selain itu, pengelasan oleh manusia mungkin tidak jim
akibat daripada tekanan dan kepenatan. Mata manusia juga menpunyai kerumitan untuk
mengasingkan warna seperti biru dan unggu. Seterusnya,manusia mempunyai persepsi yang
berlainan terhadap warna. Bagi menjayakan sistem ini, lima puluh lima limau nipis telah
digunakan untuk menjalankan eksperimen bagi mendapatkan hasil nilai gred untuk
menganggarkan tarikh luput limau nipis berdasarkan warna. Sistem mi akan meggunakan sebuah
kamera dan menangkap gambar limau nipis dalam format imej warna RGB. Selepas itu, latar
belakang imej RGB akan dikeluarkan dengan menggunakan teknik menapis median. Akhirnya,
peratusan wama hijau dikeluarkan daripada gambar. Hasil nilai tersebut digunakan untuk
menentukan gred dan anggaran tarikh lupusan limau nipis. 1-Tasil terakhir daripada ujian sistem
telah memperolehi keputusan seperti mi 90% lulus gred A, 25 % lulus gred B, 71.43 % lulus
gred C,93.75% lulus gred D dan peratus kelulusan keseluruhan ialah 80.39%..
vii'
TABLE OF CONTENTS
CHAPTER TITLE
2
PAGE
DECLARATION
ii
SUPERVISIOR'S DECLARATION
iii
DEDICATION
iv
ACKNOWLEDGEMENT
v
ABSTRACT
vi
ABSTRAK
vii
TABLE OF CONTENTS
viii
LIST OF TABLES
xi
LIST OF FIGURES
xii
LIST OF APPENDICES
xiv
INTRODUCTION
1.1 Background
1.2 Problem Statement
1.3 Objectives
1.4 Scopes
1.5 Significance of the study
1.6 Thesis Organization
1
1
2
3
4
4
LITERATURE REVIEW
2.1 Introduction
2.2 Overview
2.3 Exiting case study
2.3.1 Color grading in tomato maturity estimator
using image processing technique
2.3.1.1 Method
2.3.1.2 Advantages and disadvantages of the
system
2.3.2 Mango grading by using fuzzy image analysis
2.3.2.1 Method
2.3.2.2 Result and Discussion
6
7
8
9
9
5
9
11
12
12
14
ix
3
2.3.2.3 Advantages and disadvantages of the system
2.3.3 Objective color measurement of tomatoes
and limes
2.3.3.1 Method
2.3.3.2 Advantages and disadvantages
2.4 Manual flow of key lime
2.5 Proposed new system
2.6 Conclusion
15
16
METHODOLOGY
22
22
22
24
25
27
28
28
29
30
33
33
34
35
3.1 Introduction
3.2 Software Development Life Cycle (SDLC)
3.2.1 Requirement planning
3.2.1.1 Image acquisition
3.2.2 User design
3.2.2.1 Image enhancement
3.2.2.2 Feature extraction
3.2.3 Construction
3.2.3.lTesting
3.2.4 Cutover
3.3 Software and hardware requirement
3.3.1 Software requirement
3.3.2 Hardware requirement
4
5
16
18
19
20
21
4.1 Required Techniques
4.2 Image Acquisition
4.3 Image Enhancement
4.3.1 RGB Format
4.3.2 Filtering
4.3.3 Threshold
4.4 Feature Extraction
36
36
37
37
37
38
39
40
IMPLEMENTATION
5.1 Introduction
5.2 Image Acquisition
5.3 Image Enhancement
5.3.1 Filtering
5.3.2 Threshold
5.4 Feature Extraction
5.4.1 Sum of Threshold Image
41
41
41
42
42
43
44
44
DESIGN
x
5.4.2 Value of green
5.4.3 Estimate Maturity of Key lime
6
RESULT AND DISCUSSION
6.1 Introduction
6.2 Final Interface of the system
6.3 Test Result
6.4 Advantages and Disadvantages
6.4.1 Advantages of the system
6.4.2 Disadvantages of the system
6.5 Achieved Objective
6.6 Constraints
6.7 Assumptions and Further Research
6.7.1 Assumptions
6.7.2 Further Research
44
45
47
47
47
48
49
49
49
50
50
51
51
51
7
CONCLUSION
53
8
REFERENCES
54
9
APPENDIX A
55
xi
LIST OF TABLES
TABLE NO. TITLE
PAGE
2.1
Key lime color descriptions
2.2
Final result of Mango grading
15
3.1
Green value according to key lime grade 29
3.2
List of software requirements
34
3.3
List of hardware requirements
35
6.1
Result from testing
48
xii
LIST OF FIGURES
FIGURE NO.
•
TITLE
PAGE
2.1
Processes of tomato maturity estimator
10
2.2
Tomato storage life time
11
2.3
Formula of mean image
12
2.4
Formula of mean skin
13
2.5
Frame work of the system
13
2.6
Result of digital image processing of mango
14
2.7
Lemon measuring device
17
2.8
Formula for pixel of color index
17
2.9
Histogram of the lime
18
2.10
Flow chart Key lime export process by manual
20
2.11
Flow chart of Key lime grading by machine
21
3.1
The RAD models
24
3.2
Pre-set up hardware
25
3.3
Image of Key limes according to its grade
27
3.4
Sample user interface in Matlab
28
3.5
Labeled Key Limes according to grade
31
3.6
Healthy Key Lime at the cut session
32
3.7
Rotten Key Lime at the cut session
33
xl"
4.1
Original image of key lime
37
4.2
RGB image
38
4.3
Filtered image
38
4.4
Grayscale image
39
4.5
Binary image
39
5.1
Hardware Setup for image acquisition process
42
5.2
Source code for filtering
43
5.3
Source code for Threshold
43
5.4
Source code Sum of Threshold Image
44
5.5
Source code percentage of the green mean
45
5.6
Source code for classification of Key lime
45
6.1
GUI output of the system
47
xiv
LIST OF APPENDICES
APPENDIX TITLE
A
Result from testing
PAGE
55
1
CHAPTER 1
INTRODUCTION
This chapter discusses the classification of Key lime base on colour image processing briefly.
This chapter contains six sub-sections: The first section discuss the background of this
project. The next section describes the problem statement of this project. The third section
discusses the objectives for the project. The fourth section discusses the scopes for the
project. The fifth section describes the results and future benefits of the project. At last not
least, sixth section contains the thesis organization.
1.1 Background
Malaysia is one of developing country in ASEAN countries. Nowadays the agriculture sector
plays more vital roles in Malaysia economic growth. In the Ninth Malaysia Plan (9MP) stated
that "New Agriculture" will be encourage the large scale farming and increase the usage of
the modern technology In the particular plan, also mentioned about the high quality
production and a wide potential of biotechnology, increased concerns with information and
communications technology (ICT). In the other hand, influence of information and
communication technologies (ICT) in agriculture sector will upgrades the economics growth
very well [1]. Generally the usage of computer technologies on agriculture sector increase the
wide courage of the agriculture based industries to world and universal markets. Computer
technologies supports agriculture sector by the system that user-friendly interfaces, well
functioned, man power saving and cost savings.
2
In this project the Key lime (botanical name: Citrus aurantfolia) was used as the sample of
the experiments. Key lime is originated from Southeast and its other names are West Indian
lime, Bartender's lime, Omani lime, or Mexican lime. In Malaysia it was called as "Limau
nipis". It is categorized in the citrus species with fruit in 2.5-5 cm in diameter size. In
addition the fruit is yellow and green. The Key lime is very small and round in shape The
Key lime a fruit that has acid but less richly flavoured then usual limes and contain less seeds.
In this business era globalization, still there is no proper digital grading and classifying
system for the Key lime in Malaysia. This causes major negative effects Key lime farmers
and entrepreneur .Improper grading system becomes a major obstacle to our local farmers
and entrepreneurs to export their yield to foreign countries. It also indirectly affects
Malaysian economics because absence of classifying system causes the Key limes not to be
exported to foreign country. So far, in Malaysia the Key limes are used to be classifying
using manual grading method which may be incorrect sometimes. At last not least, by using
image processing techniques from ICT applications can come out with a well-designed
system to classify and grading the Key lime. On the other hand, this system not only
increases the local farmers and entrepreneurs income it may raise our government's income
by export to abroad.
1.2 Problem Statement
In Malaysian agriculture field there is some specific grading system for Key lime quality. So
far, the grading was conducted by manually by using human common sense. The manual
grading was based on the human classification on key limes colour. If the colour, of the key
lime green then it classify as grade A while its turn yellow then classify as less quality. This
method affect the key limes market prices without any symptoms. This is a one of the reason
that Key limes are unable to be exported to foreign country when though our country
produces yield in large quantities. Grading system using a digital based system is a more
needed when a yield is exporting to foreign countries.
3
There are lots of the disadvantages in the exportation or local market when the Key lime
classify in manual process. The main reason is manual method consuming man power and
time. This is because the farmers or entrepreneurs have to identify and classify the each key
lime by manually after it's plucked from tree. In addition, the physical appearance of Key
lime which is small size (2.5-5.0 cm in diameter) and round in shape causes time consuming
until its takes long time to market sells. At last not least, the grading by humans can causes
incorrect classification of Key lime. This is because there is no one is perfect and each human
beings makes mistakes. So that, there is a high probability to the farmers and farm workers
make the grading incorrectly due to their less concentration while grading by manually. Last
not least, the manual grading by physical colour of the Key lime is not a universal standard
accusation grading. As general facts, a natural human sight is very limited in differentiating
the colours such as blue, sky blue, and violet which are lies in similar range of colour.
1.3 Objective
There are some objectives to be achieves in the project:
I.
To develop a prototype to classify the Key lime maturity base on colour.
II.
To estimate the expired date of Key lime by colour.
III.
To identify the condition of the Key Lime.
4
1.4 Scope
I.
The scope of this project is owners of manufacturing factory of Key Lime and Key
Lime entrepreneurs in Malaysia.
II.
This System will be developing in Matlab by using image processing techniques based
on Key lime colour.
III.
The camera was placed in the 16cm distance from the Key Lime.
IV.
The LED lamp used to provide a constant light source to capture the Key Lime
images.
V.
The 55 Key Lime samples were used to conduct the storage lifespan estimation
experiment.
VI.
The size of the Key Lime considered as limitation in this project. The small sized Key
Lime is not considered in this project. The diameter of the key lime should be in 3cm
to 4cm.
1.5 Significance of the study
This project will improves agriculture sector to increase Key lime yield. This will helps Key
lime farmers and entrepreneurs' increases profits by classification grad system based on
colour In addition, classified Key lime can be sell with different prices in market and may
export to the foreign country. There is no any objection that the present of this system in
agriculture field especially the farmers and entrepreneurs gains benefits in the way saving
man power and time by the classification process.
5
1.6 Thesis Organization
This thesis contains seven chapters. First chapter is introduction and its included background
of the study, definition of term and problem statement. In addition, the objective, scope of
study and significance of the study also be inserted in the first chapter. The background of the
project discusses the basic details of Key lime. Other than that, the sub-chapter called
definition of terms defines abbreviation of term that used throughout this thesis. The problem
statement explains the possible problems that should be overcome in this project and the
problem major crisis faced by Key lime production sector. The objective discusses the main
aim of this project. On the other hand, the scope of this project discusses the main restriction
conditions and required characterizes of Key lime to conduct experiments by using this
system. At last not least, the significance of the study discusses the project impacts in Key
lime industry.
6
CHAPTER 2
LITERATURE REVIEW
This chapter two discusses the literature review on existing systems and techniques similar to
classification of key lime base on colour image processing. This chapter contents five
sections: The first part discuss more briefly about key lime and its benefits. The second part,
elaborate the techniques on fruit colour differentiation. In the third part, it discusses the
review on existing similar systems. The fourth part, briefly explain the current system and
techniques in local industry. In the fifth part it introduces a latest or new system of
classification of key lime to the local industry. At last not least, the conclusion holds the
summary of this chapter two.
7
2.1 Introduction
The Key lime (botanical name: Citrus aurantfolia) is a common lime species used in
Malaysia. In Malaysia, the key lime is using in large quantity by local people for many
purposes such as health, beauty concerns, medicine and etc. There are a lot of benefits from
the Key lime for the consumers in many angles. The first benefits are Key lime is traditional
medicine for coughing [2]. Second is Key lime plays a vital role in providing the in Vitamin C
for human body. Generally, the vitamin C is an antioxidant which protects our cells from
damage inflammation and keeps our immunity strong in the body [3]. In addition the Key
lime also contains limonin, a compound that helps and prevents cancer in skin, mouth,
stomach, lung and colon [3]. Next is Key Limes contain bioflavanoids a natural
antihistamines helps humans to from the relieve allergy symptoms. It also can be used as oil
controller for oily faces and blackheads remover for skins [3].
Generally, computer vision system by using image processing techniques can be used to
differentiate the fruits based on colour, size, shape and texture are considered as major
functions in the food industry [4]. In addition, the colour and size of the object or fruits are
the very important quality of natural image. Moreover it also used be performs the vital roles
in visual perception [4] .At last not least, this chapter going to discuss some existing system
and there are two selected systems similar to this title which is classification of Key lime
based on colour.
a) Colour Grading in Tomato Maturity Estimator using Image Processing Technique [7]
b) Mango Grading By Using Fuzzy Image Analysis[4]
c) ARM Based Fruit Grading and Management System Using Image Processing[8]
8
2.2 Overview
The colour is an important resource in process of identifies the ripening status and grade.
Moreover, colour represents the maturity and ripening stage of a particular fruit or vegetable.
Generally the human are responsible to identify the grades of the fruits with their naked eyes.
There are some negative impact caused by human eye prediction the fruit grade based on
colour such mild changes in colour of fruit, illumination, angle of sight. Therefore human
facings difficulties in producing very accurate procedure for visual inspection of fruits based
on colour [5].
In general, the limes dark green when ripped and when over ripped turns yellow. The
nutrition contents in lime increases with it ripening stage [6].When the lime ripe fully it's
bring more nutrition than under ripening stage. Basically, all limes begin with green and its
turn yellow when over ripe. When the limes are fully matured its will increase in size and its
external skin will be stronger [6]. At last not least, the key limes that send to international and
local market should meet some requirements based on size, colour, hygienic and healthy
fruits.
Table 2.1 Key lime colour descriptions [13]
Colour Range
Description
Black and green
Before harvesting, unripe
Dark Green
Ready for harvesting, 100% green
Greenish Yellow
80% Green, 20% yellow
Dark Yellow
100% Green
9
2.3 Existing case study
2.3.1 Colour grading in tomato maturity estimator using image processing technique [7]
This case study was conducted by W. Md. Syabrir, A. Suryanti, C. Connsynn from Faculty of
Computer Science and Software Engineering, University Malaysia Pahang in year 2008. The
main objective of this study is especially to improve the colour grading system of tomato in
the Malaysia's tomato industry. The method discussed in this case study is improving the
process of manual grading of the tomato into the Information and communication
technologies (ICT). This section will discuss the method and advantages and disadvantages
of the system of this case study.
2.3.1.1 Method
Colour grading in tomato maturity estimator was developed by using some of the image
processing techniques such as image acquisition, image enhancement and feature extraction.
In figure 2.1 the flow chart of the colour grading in tomato maturity estimator using image
processing technique has been shown.
10
I BEGIN
I
Input
Image
Acquisition
-
I i linage filtering
2. Threshold process
Image
Enhancement
Feature
Extraction
'-S
J I. Extract th
interested colorarea
Output
—
p-
Figure 2.1
Process of tomato maturity estimator
In this case study, about 50 tomatoes were used as sample in the first stage. The 50 tomato
samples were had passed some particular prototype of that system that has produced output.
Furthermore, particular prototypes are the image acquisition which is converting normal
images into grey scale images, thresholds, filtering, and features extractions.
The first step is an image acquisition. In this step the images are captured by a personal
computer's camera which was placed approximately 100mm on the top of the tomato. To
complete this step there is same background and same visible light condition provided. The
second step is an image enhancement used to indicate the required features of interest in an
image. In addition, in this system the background of the required image has been considered
as noise. So that, the removing background noise is very important in this system. On the
other hand, the image filtering and threshold are two types of image enhancement technique
in Spatial Domain methods were practiced in this system to remove the background noise.
11
The third step was features extraction. The main purpose of this step is to get the details of
the interested area in the image. For an example, a* values of the sample tomato's colour is
the interested. Next is step is background removal of the influence of the image. After that,
the total of a* values are gathered and totalled up. Based on the summation of total a* values
used to calculate the mean of a* values. In this step, boundaries tracing, removing
background and obtaining a* values are done. At last not least, the expired date of the sample
tomatoes is collected based on the storage life shown in figure 2.2.
Storage life:
Breakers (10-20% of full maturity) ........ 21 to 28 days
Turning (30-40% of full maturity) ............ 15 to 20 days
Pink (50-60%of full matwity) ................ .ltol4days
Light red (70 - 80% of full maturity) ....... 5 to 6 days
Red (full maturity) ..................................... to4days
Figure 2.2 The tomato storage life time [7]
2.3.1.2 Advantages and disadvantages of the system
The prototypes used in this system are able to estimate the expiry date of the tomatoes. Even
though the tomatoes are not available yet in the export market process but able to predict it
early. In addition, this system is helping the tomato industry by increase the product of
tomato in market. More than that, this system prototype provides a solution to the manual
estimation by manpower in determining tomato maturity. At the same time it reduces errors
in grading by using computer systems.
12
There are some disadvantages of colour grading in tomato maturity estimator system using
image processing technique is the system programmed to estimate only one single tomato for
each attempt. So that, it will consume too long time to estimate a large amount of tomatoes. It
will affect the export process of these tomatoes in local or internationally.
2.3.2 Mango Grading By Using Fuzzy Image Analysis [4]
This case study was conducted by the Tajul Rosli Razak, Mahmod B. Othman, Mohd Nazari
bin Abu Bakar, Khairul Adilahbt Ahmad, AB Razak Mansor from Universiti Teknologi Mara
(Perlis). The vital purpose of this study is on a design and development of an algorithm that
detecting and sorting the mango minimum eighty percent accuracy in grading compared to
human expert sorting manually. In addition, the main objectives of this study mango grading
process by the fuzzy image clustering for local mango in Perlis. There are also some other
objectives in this case study which are to develop fuzzy image clustering algorithm and to
compare the experimental results with manual grading system by human.
2.3.2.1 Method
This case study proposes a mango grading method for mangoes quality classification by
using fuzzy image analysis. The algorithm used to develop the system as below. The first step
is determining the size of mango by calculating the area of image mango. Secondly, detect
the colour of mango by determine the mean of three colour of red, green and blue (RGB).
Mean image = (Red value (Find size image) + Green value (Find size image) + Blue value (Find size image))/3
Figure 2.3 Formula of mean image
13
Third step is a process to apply edge detection algorithm. The main purpose of this step is to
determine skin of image mango by calculating the mean skin.
Mean skin = (edge (Red value) + edge (Green value) + edge (Blue value))/3
Figure 2.4 Formula of mean skin
The fourth step is fuzzy Inference Rule to get three values of size, colour and skin to identify
the grade of mango. At last not least, the final step is mango ranking based on its quality.
iliar
I
Pocssng
(MATLAB)
I -----
iFar
Figure 2.5 Frame work of the system
14
2.3.2.2 Result and Discussion
This system was developed in Matlab by combining the digital image processing and fuzzy
classification. The digital image processing in Matlab mainly used to extract the parameter of
the mango which is size and colour to prepare the input for fuzzy classification.
Figure 2.6 Result of digital image processing of mango
The result from the digital image processing earlier was used in fuzzy logic inference in
second part to get the final classification of mango based on colour and skin. The final
classification result is based on the fuzzy logic inferences.