2017 2nd International Conference for Convergence in Technology (I2CT)

Detection of Plant Disease Using Threshold, K- Mean

Cluster and ANN Algorithm
Trimi Neha Tete Sushma Kamlu
Department of Electrical & Electronics Department of Electrical & Electronics
Birla Institute of Technology Mesra Birla Institute of Technology Mesra
Ranchi, Jharkhand Ranchi, Jharkhand
(trimitete@gmail.com) (sskadwane@gmail.com)

Abstract- Indian economy highly depends on agricultural not beneficial to supervise the large crop fields. By only
productivity. An important role is played by the detection of watching the manisfestation on the plant leaves.This process
disease to obtain a perfect results in agriculture, and it is provides robotic guidence, automatic process control and
natural to have disease in plants. Proper care should be taken inspection to support the machine vision.When just visual
in this area for product quality and quantity. To reduce the
ways is applied for the identification of plant diseases a
large amount of monitoring in field automatic detection
techniques can be used. This paper discuss different processes large amount of labourious task is needed and at the same
for segmentation technique which can be applied for different time the detection will be inaccurate and it can only be done
lesion disease detection. Thresholding and K- means cluster in some part of areas.If the image processing is applied for
algorithms are done to detect different diseases in plant leaf. the automatic disease detection then low amount of effort
will be needed, it will have low cost an on the other side it
will be less time consuming and will be more accurate.
Keywords: Image processing, Thresholding algorithm, K-Means
cluster.
Some of the general diseases are found as black spot, cirtus
canker, early scotch, late scotch and other bacterial,viral and
fungul diseases in the plants. Image processing can be used
I. INTRODUCTION to measure the influenced area of the disease and to figure
out the differentiation between the colour of the influenced
area. A procedure of seperating or distributing a picture into
A simple method like naked eye observation is done by the different partition is called Image segmentation. In this
experts for the detection of disease in plants. When this procedure these pictures normally approach to something
process is applied then a large amount of workers are that human eye can quickly distribute and view as individual
needed to monitor if the field is very large then a very high object. Computers do not posses the ability to recognise the
amount of expenditure is also needed. And in some areas the objects so that the different techniques of the segmentation
farmers are not able to get the proper facilities or proper can be adapted. Segmentation procedure is established on
ideas due to which they are not able to get in contact with different appearence found in a picture such as colour
orientation, texture, boundaries etc.
the experts then they ask for a high amount of money and it
is also time consuming. In such condition the techniques are

II. LITRATURE REVIEW

There are few reported study to solves and describes advances utilized in agriculture field. One of the important techniques is
based on image processing. Table 1 presenting some reviewed papers along with the details of the research work and
methodology used.

Year of
S.
Author name Publicat Methodology Proposed work
no.
ion
1. Hossein Aghighi
2. Ali Akbar Matkan Machine Learning This paper discussed on machine learning
1.
3. Davoud Ashourloo 2016 Regression Techniques techniques for rust disease detection. [2]

1. Shyu Chi-Ren Image Analysis for Mapping This article discussed on image analysis in
2. 2. Jason M. Green 2007 Immeasurable Phenotypes maize (corn) plant. [3]

978-1-5090-4307-1/17/$31.00 ©2017 IEEE 523

 3. Daniel PK Lun

1. Wenjiang Huang
This work discussed importance of disease
2. Qingsong Guan
detection on crops and developed New
3. 3. Juhua Luo New Spectral Indices (NSIs)
2014 Spectral Indices (NSIs) for the reason.
1. Anand R
K-means clustering This work presents a method for identifying
2. Veni S
algorithm and Artificial plant leaf disease using segmentation and
4. 3. Aravinth J 2016 Neural Network classification techniques for brinjal leaf. [5]

K-means clustering, Otsu

It is presenting the different methodology
1. Sachin D. Khirade Threshold Algorithm,
used for the identification of plant diseases
Color co-occurrence
2. A. B. Patil using their lesions image along with
5. Method,
2015 algorithms for segmentation and feature
Back Propagation Neural
extraction .[6]
Network Algorithm
1. Rahat Yasir
2. Md. Ashiqur Rahman Computer vision based techniques has been
6. Artificial Neural Network
3. Nova Ahmed proposed for skin diseases. [7]
2014

1. Muhammad Thaqif
Bin New approach for an orchid disease
Image processing and Fuzzy detection with the help of Image
7. 2. Mohamad Azmi
logic Processing and Fuzzy Logic methodologies
3. Naimah Mat Isa 2013
has been explained.[8]

1. Pranjali B. Padol This paper is proposed classification and

8. 2. Prof. Anjali A. Yadav recognition of grape leaf diseases using
2016 SVM Classifier
SVM technique.[9]
1. Manjula Sri Rayudu Nearest Neighborhood (NN)
The paper describes the automatic detection
2. Vaibhav Jain classifier, Maximum mean
9. of eye diseases using Image Processing
3. M. Rao Kunda distance classifier and K-NN
2012 techniques using various classifiers.[10]
classifier
1. K.Bhima This work discussed about brain tumor
Marker based watershed
10. 2. A.Jagan detection using marker based watershed
2016 technique
technique. [11]

III. METHODOLOGY
The methodology is used in this paper are two different
segmentation techniques such as thresholding and K-means Pre processing of sample
clustering algorithm. First the digital images of plant leave
are acquired form field using camera. Then image pre
processing is done. After that two different segmentation
Segmentation of sample
processes are done to segment the original image and extract
useful features to identify the infected parts of the plant leaf.
Image processing block diagram is shown in Figure 1.
Feature extraction of sample

Classification of sample

Figure 1: Block diagram for Image

Processing

524
 Start

Read image

Convert image from RGB to L*a*b color space

Early scorch Citrus canker L=0.2126*R+0.7152*G+0.0722*B
a=1.4749*(0.2213*R-0.3390*G+0.1177*B)+128
b=0.6245*(0.1949*R+0.6057*G-0.8006*B)+128

Classify the colors in a*b space using K-

means clustering

Plot every pixel in the image using the results

Black spot Late scorch from K-means

Figure 2: Sample images of Infected Leaves

Masking red, green and blue pixels

Stop
The proposed approach of the image segmentation
such as thresholding and K- means cluster algorithms are Figure 4: Flowchart for K-means cluster
illustrated in flowchart below. The RGB image of infected
leaves was picked up. Figure 2 shows the sample of infected IV. RESULTS
leaves. It is obvious from Figure 2 shows the different types
of diseases in plant leaves such as early scorch in
Chrysanthemum (guldaudi) leaf, citrus canker in lemon leaf,
black spot in rose leaf and late scorch in bitter gourd leaf.

Start

Threshold Value:95
Fig.(a) Chrysanthemum
Disease Name: Early
(guldaudi) leaf
Read image scorch

If
Yes RGB image<=Threshold No Threshold Value:100
value Fig.(b) Lemon leaf Disease Name: Citrus
canker
Change pixel Change pixel
value of RGB value of RGB
image to zero image to
(Black) 255(white)

Show the output Threshold Value:140

Fig.(c) Rose leaf
Disease Name: Black spot

Stop

Figure 3: Flowchart for Thresholding algorithm Threshold Value:50

Fig.(d) Bitter gourd leaf
Disease Name: Late scorch

525
 V. CONCLUSION

This paper discussed various techniques of segmentation for

identification of different the plant disease. This paper also
discussed classification technique to classify the plant
diseases. For the various input, k- Figures show the original
images which are followed by output of thresholding and K-
RGB means cluster for segmented image. Though, algorithm of k
Cluster 1 Cluster 2 Cluster 3 Blue mask
image
mean clustering requires a priori specification about number
of cluster centers this technique gives relatively effective
result as compared to the thresholding and gives best result
for distinct data set.
RGB
Cluster 1 Cluster 2 Cluster 3 Blue mask
image
VI. REFERENCE

[1] Singh Vijai and A. K. Mishra “Detection of unhealthy region of plant

leaves using image processing and genetic algorithm.” Computer
Engineering and Applications (ICACEA) International Conference on
RGB Advances in IEEE, 2015.
Cluster 1 Cluster 2 Cluster 3 Blue mask
image [2] Ashourloo, Davoud, Hossein Aghighi, Ali Akbar Matkan,
Mohammad Reza Mobasheri, and Amir Moeini Rad, “An
Investigation Into Machine Learning Regression Techniques for the
Leaf Rust Disease Detection Using Hyperspectral Measurement”,
2016.
RGB [3] Shyu, C. R., Green, J. M., Lun, D. P., Kazic, T., Schaeffer, M. and
Cluster 1 Cluster 2 Cluster 3 Blue mask Coe. E. “Image Analysis for Mapping Immeasurable Phenotypes in
image
Maize [Life Sciences]”, IEEE Signal Processing Magazine, 24(3),
Figure 6: Output of K-means cluster algorithm 115-118. 2007.
[4] Huang, Wenjiang, Qingsong Guan, Juhua Luo, Jingcheng Zhang,
Jinling Zhao, Dong Liang, Linsheng Huang, and Dongyan Zhang.
“New optimized spectral indices for identifying and monitoring
winter wheat diseases.” IEEE Journal of Selected Topics in Applied
Earth Observations and Remote Sensing 7, no. 6: 2516-2524, 2014.
Lemon [5] Anand, R., S. Veni, and J. Aravinth. “An application of image
processing techniques for detection of diseases on brinjal leaves using
Bitter gourd
k-means clustering method.” Recent Trends in Information
100 Rose Technology (ICRTIT),International Conference on. IEEE, 2016.
Chrysanthemum [6] Khirade, Sachin D., and A. B. Patil. “Plant Disease Detection Using
90 Image Processing” Computing Communication Control and
Automation (ICCUBEA), International Conference on. IEEE,2015.
80 [7] Yasir, Rahat, Md Ashiqur Rahman and Nova Ahmed
“Dermatological disease detection using image processing and
70 artificial neural network” Electrical and Computer Engineering
(ICECE), International Conference on. IEEE, 2014.
60 [8] Bin Mohamad Azmi, Muhammad Thaqif and Naimah Mat Isa,
“Orchid disease detection using image processing and fuzzy
50 logic.” Electrical Electronics and System Engineering (ICEESE),
International Conference on. IEEE, 2013.
[9] Padol, Pranjali B. and Anjali A. Yadav. “SVM classifier based grape
40 leaf disease detection”. In Advances in Signal Processing (CASP),
Conference on pp. 175-179. IEEE, 2016
30 [10] Ravudu, ManiuleSri, Vaibhav Jain and M. M. Rao Kunda, “Review
of image processing techniques for automatic detection of eye
20 disease” Sensing Technology (ICST), Sixth International Conference
on. IEEE, 2012.
10 [11] Bhima, K. and A. Jagan “Analysis of MRI based brain tumor
identification using segmentation technique.” Communication and
0 Signal Processing (ICCSP), International Conference on IEEE, 2016.
2 4 8 11 12 15 16
Number of cluster

Figure 7: Distance metric of samples

526