See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/339363157

Identification Of Visually Similar Vegetable Seeds Using Image Processing And

Fuzzy Logic

Article  in  Journal of Scientific Research and Development · February 2020

CITATIONS READS

7 1,196

3 authors, including:

Edwin Arboleda Elbert Manangan Galas

Cavite State University, Indang, Cavite,Philippines Pangasinan State University
41 PUBLICATIONS   274 CITATIONS    9 PUBLICATIONS   32 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Fuzzy Logic and Image Processing View project

Image Processing View project

All content following this page was uploaded by Edwin Arboleda on 19 February 2020.

The user has requested enhancement of the downloaded file.

 INTERNATIONAL JOURNAL OF SCIENTIFIC & TECHNOLOGY RESEARCH VOLUME 9, ISSUE 02, FEBRUARY 2020 ISSN 2277-8616

Identification Of Visually Similar Vegetable Seeds

Using Image Processing And Fuzzy Logic
John Andrew A. Garcia, Edwin R. Arboleda, Elbert M. Galas

Abstract: Computer vision and image processing have attracted more and more interest of researchers because of its wide applications in many fields.
With this in consideration, this paper conducted a study comparison and identification of two vegetable seeds through fuzzy logic and image processing.
Images taken were processed and three morphological features were used as rules for the fuzzy inference system. The result was that the fuzzy
inference system worked as expected even if there are only two true values for each characteristic.

Index Terms: Fuzzy Logic, Image Processing, Morphological Feature, Vegetable Seeds
——————————  ——————————

1. INTRODUCTION several areas. Some of them are manufacturing product

THE execution of computer vision (CV) automation in line with review, traffic observation, leisure, and medicinal intervention.
taking, processing, and evaluating pictures can be efficiently Aside from that, it has been effectively employed in agricultural
applied in the nondestructive analysis of visual quality features manufacture in several tasks like automated evaluating,
in agricultural crops' seeds [1]. Image processing, also termed reaping, and classifying of goods such as food, fruit,
as digital image processing, is described by the encompassing vegetables or plant sorting [1]. Vegetable's enlarged
procedures of the principal and distinguished method of a manufacture entails the class of vegetable seeds and the
computer vision system. A two-dimensional (2D) picture serves validity of assortment as its vital considerations [5]. Clarity,
as the load of the scheme in the procedure. It generates an propagation, and free from any ailments are the initial
improved picture otherwise a group of significant variables foundation of modifying the quality of the kernels [6]. The
associated with the picture [2]. Smart classifiers can memorize significant tasks in the agricultural area are the examination
and perform numerous functions such as categorization or and sorting of the seeds. These tasks give a further worth in
sorting [3]. The examination of the seed picture contains the the manufacture, quality development, class management,
strategies of image attainment, preprocessing, and and degradation determination of vegetable kernels. These
understanding of which fallouts in assessment and sorting of activities are commonly performed by experts who manually
seeds. It can deliver supplementary information in the seed's examine and check every seed sample. Agricultural crops
manufacture, seed class management, as well as the such as grains and vegetable seeds are of dissimilar forms
recognition of seed degradation [4]. Computer vision and and tiny dimensions [7]. Therefore, this complex procedure is
image analysis are a non-destructive and profitable method for heavy, tiresome and consumes a lot of time [6][8]. Machine
categorization and classifying of agricultural and food crops vision is a controllable and exceptionally intelligent technology
throughout supervision processes and trade basis. The being widely used [9]. Machine vision was also utilized for
variation of application in agricultural and food crops is discriminating different varieties of seeds and for distinguishing
associated with the diverse tactics founded on image a type of seed from another type or for identifying damaged
examination and processing. These methods have involved kernels in wheat using a color machine vision system. It was
greater attention from researchers due to its extensive use in also used in the quality assessment of fodder [10].
Several computer-aided machine vision systems, that extensively applied in evaluating automated grading,
mechanically review and quantitatively measure grains, have distinguishing, and investigation of goods excellence in food
advanced computer knowledge, especially in artificial manufacturing [12][14]. To take out these characteristics from
intelligence. Data gathering, morphological characteristic the pictures of the seed sample, various image processing
extrication as well as their illustration, classifier/algorithm sets of rules are accessed. This lets the machine vision
option and knowledge, and classifier testing are the utmost appropriate for such activity [15]. In this case, the undesirable
vital procedures [12]. Prior to that, the examination of a digital influence of the temperature, moisture, light, etc. can alter
image gives an intention and precision measurable system for seed morphology [3].
the approximation of morphological characteristics [13]. been widely established [11]. These systems use computer
Morphological features can be identified as shape, size, color, vision technologies including several stages, which
orientation, roundness, dimension, and compactness are necessitate This study aims to identify whether the seed is a
———————————————— cauliflower or pechay (Chinese cabbage) through image
 John Andrew A. Garcia is from the Department of Computer and processing and fuzzy logic. The morphological features of the
Electronics Engineering, College of Engineering and Information image and the fuzzy inference system will all be determined
Technology,Cavite State University, Indang, Cavite. using the MATLAB program. This is to aid people that are
 Edwin R. Arboleda is from the Department of Computer and having a hard time identifying the type of seed since they look
Electronics Engineering, College of Engineering and Information
Technology,Cavite State University, Indang, Cavite. E-mail:
visually similar in the naked eye.
[email protected]
 Elbert M. Galas is from the Information Technology Department, 2 RESEARCH METHOD
College of Computing, Pangasinan State University-Urdaneta
Campus
2.1 Image Acquisition and Processing
The seeds were acquired at Aguado, Trece Martires City, from
the relatives of the author. A total of 160 seeds were used in
4925
IJSTR©2020
www.ijstr.org
 INTERNATIONAL JOURNAL OF SCIENTIFIC & TECHNOLOGY RESEARCH VOLUME 9, ISSUE 02, FEBRUARY 2020 ISSN 2277-8616

this paper: 60 pechay and 60 cauliflower seeds as training seed, depending on the morphological data and the rules.
data and 20 for each seed too were used as test samples. The Sugeno type of inference system was used since the outputs
images for the samples were taken using a Canon PowerShot were constant.
SD1300IS at 12M (4000x3000) resolution with a tripod to
prevent unnecessary movements or shaking while taking the 3 RESULTS AND ANALYSIS
images. The flash of the camera was turned on for every
image to ensure proper lighting for the samples. The seeds 3.1 Image Processing
were placed in a way that it won't touch each so that the image The image taken in RGB format shown in figure 2 underwent
will give accurate data for processing. The lenses of the several processes. The first process was to format the raw
camera have a distance of 8cm for every sample image for image into a grayscale format. The image was then
uniformity. Since the images were taken uniformly, resizing of complemented and converted into a black and white format in
images is not required anymore. The images of the samples figure 3. This turns the white areas unoccupied by the seeds
taken were processed using MATLAB R2015a installed in into while the seeds itself becomes white. This makes it easier
Lenovo G41 Laptop with a processor of AMD A8 quad-core for the special function vislabels to label the seeds that were
with speeds up to 2.1GHz and has a RAM of 4GB. This laptop identified in the image.
uses Windows 10 SL operating at x64 platform. The
percentage amongst image sections and diversities of
dissimilar shades RGB histogram are then considered [16].

2.2 Morphological Feature

Three morphological structures of the model seeds were
extricated [17][18]. These are its area, perimeter, and
diameter. These features were extracted with the use of
MATLAB. MATLAB is one of the simplest and most efficient
software where image processing can be completed. Things
can be identified and sorted, defects can be determined, and
more other applications can be done in applying image
processing. Also, this software is an efficient instrument in Figure 2. Original Image
digital image processing since it can examine the RGB values
for every pixel [12].

2.3 Fuzzy Inference System

A fuzzy system is implemented in the identification of the two
seeds. It is known as a mathematical system that examines
input values that are analog based on the variables from the
values between 0 and 1. It can create intelligible and user-
friendly models of real-world systems [19][20]. A control
system used for controlling the fuzzy logic-based system is
known as the fuzzy controller. It comes from the set of fuzzy
rules that shows the behavior of the system [21][22]. An
example of a fuzzy controller is shown in Figure 1.

Figure 3. Black and White Image

3.2 Morphological Feature Analysis

Three (3) morphological features of the seeds were measured
in this paper. This is the seeds' area, perimeter, and diameter.
First, after the image was processed into a black and white
format, the holes are filled. The next step is to apply the
vislabels function so that the seeds in the image can be
labeled and distinguished from the noise of the image. Figure
4 shows the image was vislabels function was used. After this,
a range of values is used for idx so that only the seeds will be
Figure 1. Fuzzy Controller left and most of the noise will be removed from the image as
can be seen on the output, figure 5. Table 1 shows the
Fuzzy inference is known as the procedure of mapping, given minimum and maximum values for the three morphological
from input to output through the use of fuzzy logic [23][24]. features of the seeds. This table will be further used for the
Some of the notable applications of these systems are fuzzy logic analysis.
automatic control, computer vision, analysis of decisions,
expert systems, and classification of data [25][26][27][28].
Fuzzy Logic was used in this paper is identifying the type of

4926
IJSTR©2020
www.ijstr.org
 INTERNATIONAL JOURNAL OF SCIENTIFIC & TECHNOLOGY RESEARCH VOLUME 9, ISSUE 02, FEBRUARY 2020 ISSN 2277-8616

Figure 4. Labeled Image

Figure 6. Fuzzy Design

Figure 5. Output Image

TABLE 1
MORPHOLOGICAL FEATURES RANGES

3.3 FUZZY INFERENCE RESULTS Figure 7. Rule Viewer

A total of fourteen (14) fuzzy rules were formulated and used.
The basis of these rules was the ranges of morphological Figure 8 shows the surface plots of the output. It can be
features. The rules were set to be as that, as long as there are inferred here that patola has the highest range since all of the
two (2) true values for the features, the output will be like the combination in X and Y axes results with patola occupying
output when all of the three values are true. The minimum most of the plots.
value rule was used in the fuzzy inference system. Figure 6
shows the fuzzy logic design. It is shown that there are three
inputs which are the morphological features. Sugeno type
inference was used since the output should be constant.
Figure 7 shows the rule viewer where it can be seen that there
are a total of fourteen rules. Here, as we move the values from
the three inputs, the output changes. The outputs were labeled
as pechay = 0 and cauliflower = 1. The values can be rounded Figure 8. Surface Plots
off to its nearest whole number in getting the results. Values of
output < 0.5 are pechay and the rest are cauliflower. Table 2 shows the results of testing 20 samples for each type
of seed. This was done by copying the test data from MS
Excel into Notepad and the proper formatting so it can be
pasted easily at the rule viewer of MATLAB. 100% accuracy
was obtained with pechay seeds sample while only 65%
accuracy was obtained with cauliflower seeds. This gives a
total of 82.5% accuracy with 40 samples combined.

TABLE 2
FUZZY TESTING RESULTS

4 CONCLUSIONS
This paper proved that the use of image processing and fuzzy
logic to determine the type of seed is effective and has much
4927
IJSTR©2020
www.ijstr.org
 INTERNATIONAL JOURNAL OF SCIENTIFIC & TECHNOLOGY RESEARCH VOLUME 9, ISSUE 02, FEBRUARY 2020 ISSN 2277-8616

fewer errors. Although the use of traditional methods of ision based Q uality A nalysis of R ice G rains,‖ 2017
classifying of seeds is still very effective, machine vision-based IEEE Int. Conf. Power, Control. Signals Instrum. Eng.,
ones can further improve the accuracy of this. With a result of pp. 1052–1055, 2017.
82.5% accuracy when the samples were combined, this paper [12] E. R. Arboleda, A. C. Fajardo, and R. P. Medina, ―An
achieved its objective. This is slightly lower than the results image processing technique for coffee black beans
obtained by using K-Nearest Neighbor classifier which yielded identification,‖ 2018 IEEE Int. Conf. Innov. Res. Dev.
an 85% accuracy although the fuzzy logic results at pechay ICIRD 2018, no. May, pp. 1–5, 2018.
test sample were more accurate. This study can be improved [13] S. R. Mokle, P. H. K. Waghmare, and P. S. Verma, ―A
by using more samples and varieties of vegetable seeds since Review Paper on Seed Quality Analysis Using image
the scope of this is very limited to two types only. Future processing,‖ Ijsart, vol. 3, no. 1, pp. 351–355, 2017.
researchers may opt to use a wider variety of seeds and [14] E. Faculty, ―Digital Image Based Identification of Rice
morphological characteristics that don't have many Variety Using Image Processing and Neural Network,‖
overlapping features. Indones. J. Electr. Eng. Comput. Sci., vol. 16, no. 1,
pp. 182–190, 2015.
REFERENCES [15] A. Arefi, A. M. Motlagh, and A. Khoshroo, ―Recognition
[1] V. G. Narendra and K. S. Hareesh, ―Prospects of of weed seed species by image processing,‖ J. Food,
Computer Vision Automated Grading and Sorting Agric. Environ., vol. 9, no. 1, pp. 379–383, 2011.
Systems in Agricultural and Food Products for Quality [16] S. Khunkhett and T. Remsungnen, ―Non-destructive
Evaluation,‖ Int. J. Comput. Appl., vol. 1, no. 4, pp. 1– identification of pure breeding Rice seed using digital
12, 2010. image analysis,‖ JICTEE 2014 - 4th Jt. Int. Conf. Inf.
[2] M. P. Raj and P. R. Swaminarayan, ―International Commun. Technol. Electron. Electr. Eng., pp. 4–7,
Journal on Recent and Innovation Trends in 2014.
Computing and Communication Applications of Image [17] K. Edve, N. T. Rabe, E. R. Arboleda, and R. M.
Processing for Grading Agriculture products,‖ pp. Dellosa, ―Fuzzy Logic Based Vehicular Congestion
1194–1201, 1194. Estimation Monitoring System Using Image Processing
[3] S. Aygun and E. O. Gunes, ―Computer vision And KNN Classifier,‖ no. August, 2019.
techniques for automatic determination of yield [18] E. R. Arboleda, ―Comparing Performances of Data
effective bad condition storage effects on various Mining Algorithms for Classification of Green Coffee
agricultural seed types,‖ 2016 5th Int. Conf. Agro- Beans,‖ no. August, 2019.
Geoinformatics, Agro-Geoinformatics 2016, pp. 1–6, [19] G. Louverdis and I. Andreadis, ―Design and
2016. Implementation of a Fuzzy Hardware Structure for
[4] N. Pandey, S. Krishna, and S. Sharma, ―Automatic Morphological Color Image Processing,‖ vol. 13, no. 3,
Seed Classification by Shape and Color Features pp. 277–288, 2003.
using Machine Vision Technology,‖ Int. J. Comput. [20] S. V. Multiobjective, ―Automatic Tuning of a Fuzzy
Appl. Technol. Res., vol. 2, no. 2, pp. 208–213, 2013. Visual System Using Evolutionary Algorithms :
[5] J. Liu, R. Yu, K. Xiong, and Z. Han, ―Study on Approaches,‖ vol. 16, no. 2, pp. 485–501, 2008.
vegetable seed electrophoresis image classification [21] E. Schafers, ―for D a ew,‖ pp. 1033–1038.
method,‖ Proc. Int. Conf. Inf. Comput. Sci. ICIC, pp. [22] L. Xiaoshen, ―Fuzzy Inference Modeling Method of
37–40, 2012. Time-varying System Based on T-S Fuzzy System,‖
[6] V. Sandeep Varma and K. K. Durga, ―Seed image vol. 2, no. 1, pp. 335–339, 2015.
analysis: its applications in seed science research,‖ Int. [23] T. Oyama, S. Tano, and T. Arnould, ―A tuning method
Res. J. Agric. Sci., vol. 1, no. 2, pp. 30–36, 2013. for fuzzy inference with fuzzy input and fuzzy output,‖
[7] L. Gao, C. Zhao, and M. Liu, ―Segmentation of pp. 876–881, 1896.
touching seeds based on shape feature and multiple [24] Z. Sun, K. Au, and T. Choi, ―A Neuro-Fuzzy Inference
concave point detection,‖ IST 2017 - IEEE Int. Conf. System Through Integration of Fuzzy Logic and
Imaging Syst. Tech. Proc., vol. 2018-Janua, pp. 1–5, Extreme Learning Machines,‖ vol. 37, no. 5, pp. 1321–
2018. 1331, 2007.
[8] O. Adjemout, K. Hammouche, and M. Diaf, ―Automatic [25] A. Of, N. N. Control, T. O. Automatic, and T. Operation,
seeds recognition by size, form and texture features,‖ ―APPLICATION OF FUZZY NEURAL NETWORK
2007 9th Int. Symp. Signal Process. its Appl. ISSPA CONTROL TO AUTOMATIC TRAIN OPERATION.‖
2007, Proc., pp. 2–5, 2007. [26] S. Taneja, ―A NEW APPROACH FOR DATA,‖ pp. 22–
[9] L. Peilin and L. Hong, ―An Automatic Sorting System 27, 2016.
for Sorting Metal Cylindrical Workpiece Based on [27] H. Li, C. Wen, R. Gu, and Y. Zhou, ―Research on
Machine Vision and PLC Technology,‖ 2017. Intelligent Decision Evaluation Method for substation
[10] M. T. Dunn, J. Billingsley, and S. Member, ―The use of Based on Fuzzy Analysis and Decision Theory.‖
machine vision for assessment of fodder quality,‖ pp. [28] O. Ri et al., ―9lvlrq %dvhg &odvvlilfdwlrq ri )uhvk )uxlwv
179–184, 2007. 8vlqj )x]]\ /rjlf,‖ pp. 3932–3936, 2016.
[11] T. G. Devi, P. Neelamegam, and S. Sudha, ―M achine V

4928
IJSTR©2020
View publication stats
www.ijstr.org