In this computing era, image processing has
spread its wings in human life upto the extent that image
has become an integral part of their life. There are various
applications of image processing in the field of commerce,
engineering, graphic design, journalism, architecture and
historical research. In this research work, Image
processing is considered for the analysis of plant leaf
diseases. Plant leaf diseases can be detected based on the
disease symptoms. Here, dataset of disease affected leaves
is considered for experimentation. This dataset contains
the plant leaves suffered from the
AlternariaAlternata,Cercospora Leaf Spot, Anthracnose
andBacterial Blight along with some healthy leaf images.
For this analysis, an autonomous approach of modified
SVM-CS is introduces. Here, concept of cuckoo search is
considered to optimize the classification parameters. These
parameters further help to find more accurate solutions.
This autonomous approach also extracts the healthy
portion and disease affected leaf portion along with the
accuracy of results.
Study on Air-Water & Water-Water Heat Exchange in a Finned Tube Exchanger
Plant Leaf Disease Analysis using Image Processing Technique with Modified SVM-CS Classifier
1. IJEMT Volume 5, Issue 1 ( JAN-MARCH, 2017) ISSN: 2320-7043) 2017
International Journal of Engineering & Management Technology [http://www.ijemt.net]
Plant Leaf Disease Analysis using Image Processing
Technique with Modified SVM-CS Classifier
Vishal Mani Tiwari&Tarun Gupta
Computer Science, RadhaGovind Group of Institutions, Meerut, India
tiwarikivisha53@gmail.com
Abstract— In this computing era, image processing has
spread its wings in human life upto the extent that image
has become an integral part of their life. There are various
applications of image processing in the field of commerce,
engineering, graphic design, journalism, architecture and
historical research. In this research work, Image
processing is considered for the analysis of plant leaf
diseases. Plant leaf diseases can be detected based on the
disease symptoms. Here, dataset of disease affected leaves
is considered for experimentation. This dataset contains
the plant leaves suffered from the
AlternariaAlternata,Cercospora Leaf Spot, Anthracnose
andBacterial Blight along with some healthy leaf images.
For this analysis, an autonomous approach of modified
SVM-CS is introduces. Here, concept of cuckoo search is
considered to optimize the classification parameters. These
parameters further help to find more accurate solutions.
This autonomous approach also extracts the healthy
portion and disease affected leaf portion along with the
accuracy of results.
Keywords— Plant Leaf Diseases,Cuckoo Search, Support
Vector Machine, Image Processing, Agriculture crops
I. INTRODUCTION
Nowadays, uncertain changes in global circumstances are
greatly affecting the environmental and weather conditions
unconditionally that directly or indirectly affect the plants and
social species. In India, agriculture is the primary occupation
of humans. But due to these unconditional weather
circumstances, plant generally gets suffered with many
bacterial/fungal/viral diseases. To control these diseases, the
initial remedy can be undertaken if the symptoms get identify
at early stage. This detection can help farmers to control the
agricultural and financial loses. Plant disease symptoms can
be retrieved using various methods to sense the disease
symptoms like Visible Spectroscopy, Light Reflectance,
Remote Sensing, Thermograph Techniques, Laser Sensing [1].
Plants can suffer from viral, fungal and bacteriadiseases. Here,
bacterial and fungal affected leaf diseases are considered for
experimentation as viruses produce no telltale signs that can
be readily observed and often easily confused with nutrient
deficiencies and herbicide injury. One method to detect these
bacterial and fungal diseases is manually detection with the
help of some botanic expert but manually detection of leaf
diseases is much laborious and time consuming task. So, there
is the need of some autonomous method to detect the plant
diseases with more efficiency as compare to manual detection.
In this paper, an autonomous image processing approach with
modified SVM-CS classifier is considered for the detection
and classification of plant leaf disease detection.Here, concept
of cuckoo search [2] is considered with support vector
machine [3]to optimize the classification parameters of SVM.
Image Processing [4] involves the steps of image
acquisition, pre-processing (Image Contrast Enhancement),
image segmentation, feature extraction, leaf disease detection
& classification. Here, image acquisition is performed by
considering RGB colour disease affected leaf image. Image
pre-processing is performed to enhance the image quality
using histogram equalization. Image segmentation is
performed using k-means clustering. Image feature extraction
is performed to extract the features of leaf disease symptoms.
Final disease analysis and classification is performed using
modified SVM-CS classifier.
In this way, plant leaf diseases can be analysed with the
help of available symptoms. Rest of the paper is organized in
the following manner. Section II describes the basic concepts
of histogram equalization, K-means clustering, SVM and CS.
Section III explains the work related to plant disease analysis.
Section IV presents the considered dataset. In section V,
proposed concept is explained. Section VI covers the results
and discussion portion and Section VII concludes the paper.
II. BASIC CONCEPTS
11
2. IJEMT Volume 5, Issue 1 ( JAN-MARCH, 2017) ISSN: 2320-7043) 2017
International Journal of Engineering & Management Technology [http://www.ijemt.net]
This section defines the basic concepts of histogram
equalization, K-means clustering, support vector machine and
cuckoo search.
A. Histogram Equalization
Here, approach of Histogram Equalization [5] is a
considered for the image enhancement. It is a traditional
approach of image contrast adjustment. In this method, image
is enhanced by the adjusting the pixel intensities. In most of
the cases, this image enhancement approach work well by
enhancing the image contrast value. In this approach, contrast
is enhanced by stretching the pixel intensity.
B. K-Means Clustering
K-means clustering [6] is an unsupervised clustering
approach. Clustering can be defined as the partitioning of
group data points into possible number of clusters. K-means
method is an unsupervised learning algorithm that has the
capability to solve various problems related to clustering. K-
Means clustering method performs the clustering of n objects
into k clusters using the nearest mean approach of each object.
C. Support Vector Machine
SVM algorithm [7] is superior learning models which are
generally associated with a learning algorithm applied to it
which analyses data, all this done to regroup or for
categorization. It also performs the non linear classifications.
It is considerable statistical approach used to resolve the
problems related to supervised regression & classification
with well-built theoretical fundamentals that follows the
principle of structural risk minimization. In the SVM, proper
selection of parameters is most important as improper
selecting of SVM parameters usually leads to very poor
generalization capabilities [8]. Searching the optimal SVM
parameters is decisive for achieving exceptional performance.
D. Cuckoo Search
Cuckoo Search (CS) is nature inspired optimization
algorithm that came under the category of Swarm Intelligence
& introduced by Yang and Deb [9]. The optimization feature
of cuckoo bird is based on the shrewd behaviour of cuckoo
bird to find its solution. Cuckoo bird works individually and
stores their egg in the nest of another bird’s nest by pursuing
their clever behaviour [10]. The way to lay the reproductive
egg in a parasitic manner is one of the important feature of
cuckoo bird. There may be chances to strike by other bird if
the host bird found the different egg in their nest, then the host
bird can destroy the egg. So, the main focus of cuckoo bird is
to find the optimized solution that can easily match their
living environment and this can be easily completed by the
notion of random walk or random walk with Lévy flight. In
the end, best optimized solution match is found as per the
problem.
III.RELATED WORK
This section presents the existing work related to plant leaf
disease analysis. The work of different authors is presented
here.
Sjadojevic et al. [11] have presented the concept of deep
convolution neural network (CNN) and fine tuning for the
identification of plant leaf diseases. Deep learning based Caffe
framework has been used along with the set of weights
learned on a very large dataset by authors. The overall results
show the efficient results for accuracy and precision
value.Mohanty et al. [12] have also used the concept of deep
convolutional neural network for the analysis of plant leaf
diseases. Deep learning based architecture of AlexNet and
GoogLeNet have been considered. Approach is limited to
applied dataset and presented approach is not able to detect
the leaf diseases if the leaf side changed apart from the front
area. Naik and Sivappagari [13] have presented the plant leaf
disease detection by incorporating the concepts of genetic
algorithm, neural network and support vector machine. For the
experimentation, authors have used the dataset of guava,
cotton, beans, moang and lemon leaf images. Accuracy is
evaluated for both of SVM and neural network.
Dandawate and Kokare[14] have used support vector
machine concept for the detection and classification of
soybean plants as diseased or healthy species. Authors have
used the SIFT approach that automatically recognizes plant
species by their leaf shape. Sannakki et al. [15]has used feed
forward back propagation Neural Network based technique
for the diagnosis and classification of diseases in grape leaf.
Further, Bhog and Pawar [16] have incorporated the
concept of neural network for the classification of cotton leaf
disease analysis. Authors have evaluated the recognition
accuracy and execution time for K-Mean Clustering method.
Ramakrishnan et al. [17]has used back propagation
algorithm for the identification of groundnut leaf diseases.
Cercospora is the common groundnut disease. Its further stage
is cercosposiumpersonatum, then phaeoisariopsis and final
stage is alternaris. This classification with the proposed
concept shows efficient results.
IV.DATASET
12
3. IJEMT Volume 5, Issue 1 ( JAN-MARCH, 2017) ISSN: 2320-7043) 2017
International Journal of Engineering & Management Technology [http://www.ijemt.net]
For the experimentation, expert dataset with the values of
plant leaf disease symptoms is considered. The proposed
integrated concept is used for the detection of plant leaf
diseases. This dataset contains the leaf images suffered with
diseases of AlternariaAlternata, Cercospora Leaf Spot,
Anthracnose and Bacterial Blight. Some sample images for
the different diseases are shown in figure 1.
Figure 1: Sample Dataset Images
V. PROPOSED ALGORITHM
Here, the proposed modified SVM-CS classifier is
presented for plant leaf disease detection and classification
using the steps of image processing. Here, concept of cuckoo
search is considered to optimize the classification parameters.
These parameters further help to find more accurate solutions.
In this proposed approach, major image processing step is
disease detection, classification & optimization which is
performed by SVM and CS. Support Vector Machine is
statistical learning concept used as the classification and
regression models. Cuckoo search can perform the local
search in the efficient manner due to presence of single
parameter apart from the population size for the optimization
of results.The detailed concept is explained as below. The
flow chart for the work is shown in figure 2.
Input: Disease affected leaf image.
Output: Leaf disease type with percentage of affected portion.
ALGORITHM
Step 1: Consider an image from the dataset of disease
affected leaves. Image is considered in RGB color value
which is further transformed in gray scale image using color
scale transformation.
Step 2:Further, image is pre-processed using histogram
equalization method for the contrast enhancement.
Step 3:The enhanced image is further segmented into
three portions using k-means clustering methodwith three
different type of Region of Interest (ROI).
Step 4: In k-means clustering, classification is based on
the minimizing the Euclidean Distance values which can be
calculated using the equation (3.1) below:
݀ሺ, ݍሻ = ݀ሺ,ݍ ሻ
= ඥሺݍଵ − ଵሻଶ + ሺݍଶ − ଶሻଶ + ⋯ + ሺݍ − ሻଶ
= ඩሺݍ − ሻଶ
ୀଵ
…Equation (1)
Step 5:Further, Region of Interest is selected from the
segmented image.
Step 6:Convert the RGB color (ROI) image into grey
scale image and maintain the Grey Level Occurrence Matrices
(GLCMs) having expert feature values.
Step 7: Extract the disease symptoms by calculating the
feature values of Skewness, Standard_Deviation,
Homogeneity, Contrast, Smoothness, Correlation, Kurtosis,
Energy, Entropy, Mean, Variance, RMS, and IDM. These are
calculated from disease affected portion.
Step 8: Apply modified SVM-CS for the classification and
analysis of disease type.
Step 9: Here, SVM is used for the feature extraction and
disease detection using the equation (2):
SVM = SVMtrain (disease_feat, disease_type)
…Equation (2)
Where,
SVMtrain is the SVM training function.
disease_feat maintains the values of disease affected
leaves of all the disease types.
disease_type maintains the corresponding disease labels
of AlternariaAlternata, Anthracnose, Bacterial Blight
and Cercospora Leaf Spot.
Step 10: Apply Cuckoo Search for the optimization and
final classification.
10.1. Initialization of number of nests and random
initial solution.
10.2. Find the Current best solution.
10.3. While (fmin> Max generation)
Get the cuckoo value by random walk
10.4.Evaluate the quality fitness Fnj.
Randomly choose nest among n, say j.
10.5. If(Fni >Fnj)
13
4. IJEMT Volume 5, Issue 1 ( JAN-MARCH, 2017) ISSN: 2320-7043) 2017
International Journal of Engineering & Management Technology [http://www.ijemt.net]
Replace j value by new solution.
End
10.6.Retain the best solution and nests.
10.7. Rank the solution and nests to choose the best.
Pass to next generation.
End while (step 10.3), else go to step 10.2.
Step 11:Finally disease type with accuracy value is
analysed and percentage of disease affected region is
evaluated by the ratio of disease data and leaf data.
14
6. IJEMT Volume 5, Issue 1 ( JAN-MARCH, 2017) ISSN: 2320-7043) 2017
International Journal of Engineering & Management Technology [http://www.ijemt.net]
Figure 2: Work Flow for Modified SVM-CS classifier
VI.RESULTS AND DISCUSSION
This section elaborates the evaluated results with proper
discussion. Windows 7 based system with 4GB of RAM,
500GB of HDD, an Intel(R) Core(TM) i7 CPU, is used for
conducting the experiments. MATLAB is used for the
simulation of work. In MATLAB, a GUI (graphical user
interface) base interface is generated for the experimentation.
The stepwise process of image processing for plant leaf
disease detection is explained in previous section. Using the
proposed modified SVM-CS classifier, experimentation is
performed for more than 150 images. From the evaluated
results, we have analysed that percentage of diseased portion
also affect the overall crops/agriculture land. Accuracy also
varies from different images. As per the expert dataset, the
detected diseases are AlternariaAlternata, Cercospora Leaf
Spot, Anthracnose and Bacterial Blight. We have also tested
the concept for the healthy diseases to analyse the accuracy of
concept. For all the images, it shows the accurate results.
There are also the chances of multiple diseases on a single leaf.
The analysed accuracy level for modified SVM-CS varies
from 96.5 % to 98.5 %. Based on this accuracy level, results
are further compared with SVM and Improved SVM [18].
This comparative analysis is shown in table 1.
Table 1: Performance Comparison
Algorithm/Classifier Accuracy Value (range in %)
SVM 65-72 %
Improved SVM 69-79 %
Modified SVM-CS 96.5-98.5 %
From the above results, we can say that proposed concept is
efficient enough for the detection of plant leaf diseases. This
comparison of results can also be show in graphical form as in
figure 3.
Figure 3:Performance comparison of Modified SVM-CS
with SVM and Improved SVM
VII. CONCLUSIONS
In this research work, we have presented image processing
approach with modified SVM-CS classifier for the detection
and classification of leaf diseases.Here, concept of cuckoo
search is integrated with SVM. CS is considered to optimize
the classification parameters. These parameters further help to
find more accurate solutions. Support Vector Machine is
statistical learning concept used as the classification and
regression models. Cuckoo search can perform the local
search in the efficient manner due to presence of single
parameter apart from the population size for the optimization
of results.As per the expert dataset, the detected diseases are
AlternariaAlternata, Cercospora Leaf Spot, Anthracnose and
Bacterial Blight. We have also tested the concept for the
healthy diseases to analyse the accuracy of concept. To test
the further optimization of results, proposed concept is
compared with SVM and Improved SVM. From the
comparison table 1 and figure 3, we can say that modified
SVM-CS is efficient enough to detect the plant leaf diseases in
accurate manner.
REFERENCES
[1]. Mahlein, Anne-Katrin, Erich-Christian Oerke, Ulrike
Steiner, and Heinz-Wilhelm Dehne. "Recent advances
in sensing plant diseases for precision crop
protection." European Journal of Plant
Pathology 133, no. 1 (2012): 197-209.
[2]. Yang, Xin-She, and Suash Deb. "Engineering
optimisation by cuckoo search." International Journal
of Mathematical Modelling and Numerical
Optimisation 1, no. 4 (2010): 330-343.
[3]. Suykens, Johan AK, and JoosVandewalle. "Least
squares support vector machine classifiers." Neural
processing letters 9, no. 3 (1999): 293-300.
[4]. Baxes, Gregory A. Digital image processing:
principles and applications. New York: Wiley, 1994.
[5]. Pizer, Stephen M., E. Philip Amburn, John D. Austin,
Robert Cromartie, Ari Geselowitz, Trey Greer, Bart
terHaarRomeny, John B. Zimmerman, and
KarelZuiderveld. "Adaptive histogram equalization
and its variations."Computer vision, graphics, and
image processing 39, no. 3 (1987): 355-368.
0
20
40
60
80
100
SVM Improved SVM Modified
SVM-CS
Chart Title
Maximum Minimum
16
7. IJEMT Volume 5, Issue 1 ( JAN-MARCH, 2017) ISSN: 2320-7043) 2017
International Journal of Engineering & Management Technology [http://www.ijemt.net]
[6]. Kanungo, Tapas, David M. Mount, Nathan S.
Netanyahu, Christine D. Piatko, Ruth Silverman, and
Angela Y. Wu. "An efficient k-means clustering
algorithm: Analysis and implementation." IEEE
transactions on pattern analysis and machine
intelligence 24, no. 7 (2002): 881-892.
[7]. Tong, Simon, and Edward Chang. "Support vector
machine active learning for image retrieval."
In Proceedings of the ninth ACM international
conference on Multimedia, pp. 107-118. ACM, 2001.
[8]. Amari, Shun-ichi, and Si Wu. "Improving support
vector machine classifiers by modifying kernel
functions." Neural Networks 12, no. 6 (1999): 783-
789.
[9]. Gandomi, Amir Hossein, Xin-She Yang, and Amir
HosseinAlavi. "Cuckoo search algorithm: a
metaheuristic approach to solve structural
optimization problems." Engineering with
computers 29, no. 1 (2013): 17-35.
[10]. Yang, Xin-She, and Suash Deb. "Cuckoo search:
recent advances and applications." Neural Computing
and Applications 24, no. 1 (2014): 169-174.
[11]. Sladojevic, Srdjan, Marko Arsenovic, AndrasAnderla,
DubravkoCulibrk, and DarkoStefanovic. "Deep
Neural Networks Based Recognition of Plant Diseases
by Leaf Image Classification." Computational
Intelligence and Neuroscience 2016 (2016).
[12]. Mohanty, Sharada P., David P. Hughes, and Marcel
Salathé. "Using Deep Learning for Image-Based Plant
Disease Detection." Frontiers in Plant Science 7
(2016).
[13]. Naik, M. Ravindra, and Chandra Mohan Reddy
Sivappagari. "Plant Leaf and Disease Detection by
Using HSV Features and SVM
Classifier." International Journal of Engineering
Science 3794 (2016).
[14]. Dandawate, Yogesh, and RadhaKokare. "An
automated approach for classification of plant diseases
towards development of futuristic Decision Support
System in Indian perspective." In Advances in
Computing, Communications and Informatics
(ICACCI), 2015 International Conference on, ISNB
no. 978-1-4799-8792-4, page no. 794-799. IEEE,
2015.
[15]. Sannakki, Sanjeev S., Vijay S. Rajpurohit, V. B.
Nargund, and ParagKulkarni. "Diagnosis and
classification of grape leaf diseases using neural
networks." In Computing, Communications and
Networking Technologies (ICCCNT), 2013 Fourth
International Conference on, ISBN no. 978-1-4799-
3926-8 , page no. 1-5. IEEE, 2013.
[16]. Bhong, Vijay S., and B. V. Pawar. "Study and
Analysis of Cotton Leaf Disease Detection Using
Image Processing." International Journal of Advanced
Research in Science, Engineering and Technology 3,
no. 2 (2016).
[17]. Ramakrishnan, M., and A. Nisha. "Groundnut leaf
disease detection and classification by using back
probagation algorithm." In Communications and
Signal Processing (ICCSP), 2015 International
Conference on, ISNB no. 978-1-4799-8081-9, page
no. 0964-0968. IEEE, 2015.
[18]. Kaur R, Kang SS. An enhancement in classifier
support vector machine to improve plant disease
detection. InMOOCs, Innovation and Technology in
Education (MITE), 2015 IEEE 3rd International
Conference on 2015 Oct 1 (pp. 135-140). IEEE.
17