IRJET- Texture based Features Approach for Crop Diseases Classification and Diagnosis-A Research

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 06 Issue: 05 | May 2019 www.irjet.net p-ISSN: 2395-0072
© 2019, IRJET | Impact Factor value: 7.211 | ISO 9001:2008 Certified Journal | Page 7469
TEXTURE BASED FEATURES APPROACH FOR CROP DISEASES
CLASSIFICATION AND DIAGNOSIS-A RESEARCH
Namrata Ghatol1, Dr. G.P. Dhok 2
1ME Student, Dept of Electronics and Telecommunication
2Associate Professor, Sipna College of Engineering and Technology, Amravati.
---------------------------------------------------------------------***----------------------------------------------------------------------
Abstract - Agriculture is the main element of economic
growth in developing countries. Plant diseases cause major
economic and production losses as well as curtailment in both
quantity and quality of agricultural production. Now a day’s,
for supervising large field of crops there has been increased
demand for plant leaf disease detection system. The critical
issue here is to monitor the health of the plants and detection
of the respective diseases. In this project, Image processing is
used to detect and classify crop diseases based on the texture
features approach. The images are taken through a high
resolution digital camera and after preprocessing, these are
then run through the various machine learning algorithms
and classified based on their color and texture features. Inthis
project basically two machine learning algorithms are used
namely K-Nearest Neighbours, Support Vector Machine. The
implementation will be done using MATLAB.
Key Words: Classification, Textures features approach,
Machine learning, K-Nearest Neighbours, Support Vector
Machine
1. INTRODUCTION
Agriculture is an integral part of the economy of a country.
Especially in developingcountrieslikeIndia..Alarge number
of factors are responsible for the contributions by the
agriculture sector to be this low like low literacy rates
among farmers, bad quality seed and availability of
resources like water. One of the most critical factors is the
diseases that the crops contract. High quality crop
production is the big challenge for farmers. . The crop
production rates are directly proportional to the each day
plant growing progress. So plant disease detection is very
important. Also the external appearance of agricultural
products is the main quality attribute. The outerappearance
greatly affects their scale value and customer’s buying
behaviour.
Therefore, disease diagnosis and correct treatment
essential for the healthy crop production process as early as
possible. The farmer’s wrong diagnosis of crop disease
causes insecticides spray inappropriately. Various image
processing techniquescanbesignificantlyappliedtoobserve
the crop growth progress and disease diagnosis.
In this project, the diseases are classified using an
image of the leaf taken by a high resolution camera. As
symptoms in most cases are noticed on the leaves, Colorand
texture features are extracted from the image and passed
through the machine learning algorithm for classification.
Machine learning-based detection and recognition of plant
diseases can provide extensive clues to identifyandtreatthe
diseases in its very early stages. Comparatively, visually or
naked eye identification of plant diseases is quite expensive,
inefficient, inaccurate and difficult. Automatic detection of
plant diseases is very important to research topic as it may
prove the benefits in monitoring large fields of crops, and
thus automatically detect the symptoms of diseases as soon
as they appear on plant leaves. This project focuses on
diseases detection and Classification of crop species like
Chilli, Soyabean, cotton, corn(maize), orange, mango,
sunflower, peanut etc. using image processing techniques.
These plants largely produced crops in India. Improving the
productivity of these crops can significantly reduce the food
deficiency and can contribute towards improvement in
health care. Hence, these crops are taken as the crops of
interest.
2. LITERATURE REVIEW
S. Arivazhagan et al [1], proposedanapproachwhere,first
conversion of an image from RGB to HSI is done and the
green pixels are masked using threshold values. The
proposed system is a software solution for detectionof plant
diseases. The developed processing scheme consists of four
main steps, first a color transformation structure for the
input RGB image is created, thenthegreenpixelsaremasked
and removed using specific threshold value followed by
segmentation process,thetexturestatisticsarecomputed for
the useful segments, finally theextractedfeaturesarepassed
through the classifier .Texture analysis is then done using
color co-occurrence matrix (SGDM). The image is then
classified using either minimum distance criterion or SVM
classifier which has 86.77% and 94.74% accuracy
respectively.
Loyce Selwyn pinto, proposed an approach [2] In which,
Image processing is used to detect and classify sunflower
crop diseases based on the image of their leaf. The images
are taken through a high resolution digital camera and after
preprocessing, are subjected tok-meansclusteringtogetthe
diseased part of the leaf. These are then run through the
various machine learning algorithms and classified based on
their color and texture features. A comparison based on
accuracy between various machine learning algorithms is

done namely K-Nearest Neighbors, Multi-Class Support
Vector Machine, Naive Bayes and Multinomial Logistic
Regression to achieve maximum accuracy. The proposed
methodology is able to classify diseases of the sunflower
crop in a very accurate and efficient way.
Manisha A. Bhange et al [3], proposed three methods for
extracting features, histogram for color, erosion concept
morphology for obtaining boundaries of the images and
color coherence vector to classify pixels. This approach
majorly consists phases namely image preprocessing,
feature extraction, clustering, training and classification.
color coherence vector features and color morphology are
used for feature extraction. K-means clustering is used for
segmentation and SVM is used for classification of the
images the accuracy is 81%. The disease considered here is
the bacterial blight of the pomegranate leaf.
It is difficult to determine the accurate disease in noisy
image. Image should be noise free for processing. Therefore,
noise reduction techniques and image enhancement are
required for desirable processing. Valliammai and
Geethaiakshmi [4] have found that the appropriate feature
extraction of leaf can be possible if input image is noise free.
The leaf vein edges not exactly visible in Gaussian noise
method. The speckle noise affected the leaf size, shape and
pattern .Therefore, Gaussian and speckle noise removal
techniques are essential to restore the noise free leafimages
for further process. These Hybrid filter method is developed
to eliminate the noise, improve the quality of image and
thereby produces better results compared to other
traditional filters.
Implementation of RGB and Gray scale images in plant
leaves disease detection –comparative study by Padmavathi
and Thangadurai [5] have given the comparative results of
RGB and Gray scale images in leaf disease finding process.In
detecting the infected leaves, color becomes an important
feature to find the disease intensity. They have considered
Gray scale and RGB images and used median filter for image
enhancement and segmentation for extraction of the
diseased portion which are used to identifythediseaselevel.
The plant disease recognition model, based on leaf image
classification, by the use of deep convolution networks have
developed. 13 kinds of diseases are identified from the
healthy leaves with the capability to differentiate leaves
from their surroundings.
Rupesh G. Mundada et al [6] have proposed an approach
where images are converted from RGB to Gray scale first,
followed by resizing and filtering them. This approach
proposes a software prototype system for early pest
detection. Images of the infected leaf are captured by a
camera and processed using image processingtechniques to
detect presence of pests. This approach is mainly used to
detect whiteflies, aphids on the affected crops at their early
stages. Feature extraction with features like contrast and
entropy is performed. Classification is done using a Support
Vector Machine.
Bindushree H B et al [7], proposedan approachwherethe
processed image is first is segmented using k means
clustering .out of the three clusters created one of the
clusters contains the disease affected area and image
features are extracted from the particular cluster usingGray
Level Co-occurrence Matrix (GLCM).Thesefeaturesarelater
fed into support vector machines (SVM).The final
classification results from SVMs indicate whether the leaf in
the image dataset is healthy or disease affected. The results
using SVM are obtained from various kernels such as linear,
polynomial, quadratic, RBF .
P. Revathi, M. Hemalatha [8] worked on classification of
diseases in cotton leaves. Authors have considered six types
of diseases in the cotton plant for classification. Based on
advanced computational techniques the significance of this
work design is to reduce the time, cost and complexity. To
identify the affected region of a leaf the author has used
Enhanced Particle Swarm Optimization (EPSO) for feature
selection. For calculating the edge,color,texturevariance for
feature analysis of the diseased part Skew divergence is
used. The result obtained using skew divergence and EPSO
technique is 98%.
Hrishikesh P. Kanjalkaret al [9], proposed the approach
where, the image is first converted from RGB to HIS.
Segmentation is done using connected componentslabeling,
thresholding is used to avoid unwanted regions. 11 features
are used in this approach and classification is done using
back propagation neural network. The accuracy of this
method is 83%.
Dheeb Al Bashish proposed an approach where [10], the
images are segmented using K-means clustering and are
then converted from RGB to HSI. The color co-occurrence
texture analysis method is used using spatial grey level
dependence matrices. Features are calculated from H and S
components. The neural network usedhereisa feedforward
back-propagation with 93% of overall success
Prof. Sanjay B. Dhaygude et al,[11], have proposed an
approach in which firstly by color transformation structure
RGB is converted into HSV space because HSVisa goodcolor
descriptor. Masking and removing of green pixels with pre-
computed threshold level. Then in the next step
segmentation is performed using 32x32 patch size and
obtained useful segments. These segments are used for
texture analysis by color co-occurrence matrix. Finally if
texture parameters are compared to texture parameters of
normal leaf

3. ARCHITECTURAL SYSTEM
For detection and diagnosis of the various crop diseases by
taking into consideration the texture and colour of leaves of
respective crops following system is used.
Fig -1: Architectural system
System’s architecture basically consists of two phases
namely learning phase and testing phase.
LEARNING PHASE-
STEP 1-DISEASE DATABASE IMAGE AND IDENTIFICATION-
In this method, diseased images of plants are captured
through the high-resolution camera to create the required
training database. Thisdatabasehasdifferenttypesofimages
which comprisesofhealthy,diseasedcropimagesandimages
are stored in jpeg format. Validation of images are done by
experts thus making the training as accurate as possible. To
collect this database we are taking help from Dr. punjabrao
DeshmukhKrishiVidyapeeth(PDKV),Amravati,Maharashtra.
STEP 2- FEATURE EXTRACTION (TEXTURE BASED)-
Feature extraction is the importantparttogracefullypredict
the infected region. In feature extraction method, various
attributes of the segmented image are extracted. Texture
oriented featuresare calculated suchascontrast,correlation,
energy, homogeneityandmean.Thesegmentedimageisthen
converted into a gray scale image. Statistical texture based
features are extracted usingGrayLevelCo-occurrenceMatrix
(GLCM). Thus the gray scale image is converted to GLCM
matrix so that the following features can be extracted.
STEP 3-CLASSIFICATION (TRAINING)-
In training phase, the classification of diseased leavesimages
into various categories is done by experts. Then the training
dataset is made and stored.
TESTING PHASE-
STEP 1-IMAGE ACQISITION-
The first step is to capture thesample of diseasedleavestobe
tested from the digital camera. The images are in RGB
form.(red, blue, green).Images are taken in controlled
environment and are stored in the JPEG format.
STEP 2-IMAGE PREPROCESSING
Image preprocessing is performed on images to highlightthe
important features of an image and make the image more
suitable foruse in particular application. The purposeofdata
preprocessing is to eliminate the noise in the image, so as to
adjust the pixel values. It contains various steps like –image
enhancement,conversionofRGBimageintograyscaleimage,
image resizing, image segmentation etc.
STEP 3-FEATURE EXTRACTION
In this step, various attributes of the segmented image are
extracted same as in testing phase above. Texture based
(GLCM). Thus the gray scale image is converted to GLCM
matrix so that texture oriented features are calculated such
as contrast, correlation, energy, homogeneity and mean.
STEP 4-DISEASE CLASSIFICATION-
In testing processes, For classification features extracted
from trainingleaves are compared with thoseextractedfrom
testing leaves. The image is then classified based on the
matched features.
STEP 5-IDENTIFICATION OF DISEASE NAME AND SUURES-
After the classification the disease namealong with remedial
measures for treatment based on the classification are
displayed to the user.
4. FEATURE EXTRACTION
Transforming the input data into the set of features is called
feature extraction. The main goal of feature extraction is to
obtain the most relevant information from the original data.
Feature extraction is the important part to gracefully predict
the infected region. In feature extraction method, various
attributes of the segmented image are extracted. Texture
oriented featuresare calculated suchascontrast,correlation,
energy, homogeneityandmean.Thesegmentedimageisthen
converted into a gray scale image .Statistical texture based
(GLCM).
Thus the gray scale image is converted to GLCM
matrix so that the following features can be extracted. Mean,
Standard Deviation and Coarseness are taken from the
segmented image before conversion.
A. Contrast
It is the measureof the intensity contrastbetweenapixeland
its neighbor over the whole image.
B. Energy
It is the sum of squared elements in the GLCM.

C. Mean
It is the average value of the elements along different
dimensions of an array.
D. Homogeneity
It is the measure of the closeness of the distribution of
elements in the GLCM to the GLCM diagonal.
E. Standard Deviation
It computes the standard deviation of the values in matrix or
array.
Where,
The goal is to generate featuresthatexhibithighinformation-
packing properties:
 Extract the information from the raw data that is
most relevant for discrimination between the
classes
 Extractfeatureswithlowwithin-classvariabilityand
high between class variability
 Discard redundant information.
5. CLASSIFICATION
In training phase, the training images of diseased leaves are
classified by experts. After classification the dataset is made
and stored. This trained dataset is further used to train the
classifier for classification of diseased leaves to be tested.
In testing processes, For classification features
extracted from training leaves are compared with those
extracted from testing leaves. The image is then classified
based on the matched features. The two Classifiers are: K-
Nearest Neighbors (KNN) and Support Vector Machine
(SVM).These techniques areselectedduetothereasonthatin
many realapplications these classifiers haveperformedwell
and also forthe fact that these classifiers differ in complexity
and speed.
A. K-Nearest Neighbors-
K-Nearest Neighbors is a simple supervised
classifying technique. The K-nearest neighbours algorithm
(kNN) is a nonparametric method used forclassification.The
input consists of the k closesttrainingexamplesinthefeature
space. In Knn classification,theoutputisaclassmembership.
An object is classified by a majority voteofitsneighbors,with
the object being assigned to the class most common among
its k nearest neighbors (k is a positive integer, typically
small). If k = 1, then the object is simply assigned to the class
of that single nearest neighbor.
kNN is a type of instance based learning, or lazy
learning. The KNN algorithm is among the simplest of all
machine learning algorithms. The training examples are
vectors in a multidimensional featurespace,eachwithaclass
label. The training phase of the algorithm consists only of
storing the feature vectors and class labels of the training
samples. In the classification phase, k is a user defined
constant, and an unlabeled vector (a query or test point) is
classified by assigning the label which is most frequent
among the k training samples nearest to that query point
B. Support vector machine (SVM)-
Support Vector Machine is a complex classifier as
compared to KNN. It was originally developed for linear
classification but later modified for multi class classification.
The Support vector machine comes in the category of
supervised learning .The SVM used for regression and
classification. But it is popularly known forclassification.Itis
a very efficient classifier. In this every object or item is
represented by a point in the n- dimensionalspace.Thevalue
of each feature is represented by the particular coordinate.
Then the items divided intoclasses by findinghyper-planeas
shown in the figure. The diagram shows supportVectorsthat
represent the coordinates of each item.
In machinelearning, support vectormachines(SVMs)are
supervised learning models with associated learning
algorithms that analyse data used for classification and
regression analysis. Given a set of training examples, each
markedas belonging to one or the other of two categories,an
SVM training algorithm builds a model that assigns new
examples to one category or the other, making it a non-
probabilistic binary linear classifier (although methods such
as Platt scaling exist to use SVM in a probabilistic
classification setting). An SVM model is a representation of
the examples as points in space,mappedsothattheexamples
of the separatecategories aredivided by a clear gap that isas
wide as possible. New examples are then mapped into that
same space and predicted to belong to a category based on
which side of the gap they fall. In addition to performing
linearclassification,SVMscanefficientlyperformanon-linear
classification using what is called the kernel trick, implicitly
mapping their inputs into high-dimensional feature spaces.

6. RESULTS AND DISCUSSION
The proposed work is implemented on Intel Core
processor i5, 4GB RAM Laptop configuration and operating
system is windows 7. MATLAB R2016a software is used to
write the programming code. In this we used Image
processing toolboxand the database of diseasedcropimages
is collected by taking help from Dr. Punjabrao Deshmukh
Krishi Vidyapeeth (PDKV),Amravati,Maharashtra.
Fig -1:System Flowchart
The feature set consisting of Contrast, Energy, Mean,
Homogeneity, Standard Deviation and Coarseness are used
and classification is done on the basis of these. The different
classifiers are comparedagainsttheeachotheronthebasisof
Accuracy. Accuracy of the classifiers have been calculated by
using the formula. We have taken samplesofeachtypeasthe
testing data. The testing images are taken under natural
settings.
Accuracy =(Numberofcorrectpredictions/Totalnumber
of samples) *100
Following tables shows the accuracy of the 2 machine
learning classifiers and shows the comparisonbetweenthem
for respective crop samples
Table -1: Accuracy of SVM and KNN for cotton and maize
Table -2: Accuracy of SVM and KNN for orange and mango
Table -3: Accuracy of SVM and KNN for sunflower and
peanut
7. CONCLUSION
An application of texture analysis in detecting and
classifying the plant leaf diseases has been explained in this
project. The image will be taken through high resolution
digital camera and in preprocessing stage the system will
perform segmentation of leaf and will analyseitwithfeature
extraction algorithms. The machine learning algorithmslike
K-Nearest Neighbors and Support Vector Machine, Random
Forest will be used for feature extraction purpose. Plant
diseases will be detected based on their texture.Various
parameters like accuracy, precision, sensitivity, error rate
will be considered . By this method , the crop disesases will
be identified at the initial stage itself and remedial measures
will be suggested to solve the respective problem .The
accuracy and low cost of the classification allow for an
effective automatic surveillance. The accuracy can be
increased by using various enhancement techniques and by
increasing the size of training dataset.

8. FUTURE SCOPE
In the future, the proposed methodology can be
integrated with other to be developed methods for disease
identification and classification using color and texture
analysis to develop an expert system for early crop foliar
disease warning and administration,where the disease type
can be identified by color and texture analysis and the
severity level estimationcanbeperformed.Theperformance
of the system can be improved in the future by using
advanced background separation methods to separate the
leaf object from a complex background. The similar
methodology can be applied to other plant foliar infections
and early warning can be provided.
REFERENCES
[1] S. Arivazhagan, R. NewlinShebiah, S. Ananthi, S. Vishnu
Varthini, “Detection of unhealthy region of plant leaves and
classification of leaf diseases using texture features”,
AgricEngInt: CIGR Journal Vol. 15, No.1, PP 211-217, March,
2013
[2]Loyce Selwyn Pinto, Argha Ray, M.Udhayeswar Reddy,
Pavithra Perumal, Aishwarya P. “Crop Disease Detection
using Texture Analysis” IEEE International Conference On
Recent Trends In Electronics Information Communication
Technology, May 20-21, 2016, India
[3] Ms. Manisha A. Bhange, Prof. H. A. Hingoliwala,
“Detection of Bacterial Blight on Pomegranate
Leaf”,International Journal onRecentandInnovationTrends
in Computing and Communication, Volume: 3 Issue: 6, ISSN:
2321-8169 3682 - 3685, PP 3682-3685, June 2015
[4] Valliammai, N.and Geethaiakshmi,S.N.,“Multiple noise
reduction using hybrid method for leaf recognition”, 2012
International Conference on Devices, Circuits and Systems
(ICDCS), 15-16 March 2012, Coimbatore, India.
[5]Padmavathi,K. and Thangadurai, K.,“Implementation of
RGB and grayscale images in plant leavesdiseasedetection –
comparative study”, Indian Journal of Science and
Technology, Vol. 9,pp. 1-6, 2016.
[6] Rupesh G. Mundada, Dr. V. V. Gohokar, “Detection and
Classification of Pests in Greenhouse Using Image
Processing”, IOSR Journal of ElectronicsandCommunication
Engineering (IOSR-JECE) e-ISSN: 2278-2834, p-ISSN: 2278-
8735.Volume 5, Issue 6, PP 57-63,Mar. - Apr. 2013
[7] Bindushree H B, Dr. Sivasankari G G, “Detection of Plant
Leaf Disease Using Image Processing Techniques”,
International Journal of Technology Enhancements and
Emerging Engineering Research, Vol 3, Issue 04, ISSN:2347-
4289, PP 125-128
[8] P.Revathi and M.Hemalatha, “Classification of Cotton
Diseases Using Cross Information Gain Minimal Resource
Allocation Network ClassifierwithEnhancedParticleSwarm
Optimization,” Journal of Theoretical and Applied
Information Technology, vol. 60, no. 1, February 2014
[9]Mr. Hrishikesh P. Kanjalkar, Prof. S.S.Lokhande,
“Detection and Classification of Plant Leaf Diseases using
ANN”, International Journal of Scientific & Engineering
Research, Volume 4, Issue 8, ISSN 2229-5518, PP 1777-
1780,August-2013
[10]Dheeb Al Bashish, Malik Braik, SuliemanBani-Ahmad, “A
Framework for DetectionandClassificationofPlantLeadand
Stem Diseases”, 2010 International ConferenceonSignal and
Image Processing,PP 113-118, IEEE
[11] Prof. Sanjay B. Dhaygude, Mr.NitinP.Kumbhar
“Agricultural plant Leaf Disease Detection Using Image
Processing”, International Journal of Advanced Research in
Electrical, Electronics and Instrumentation EngineeringVol.
2, Issue 1,PP 599-602, ISSN: 2278 – 8875,, January 2013
[12] Al-Bashish, D., M. Braik, and S. Bani-Ahmad. 2011.
Detection and classification of leaf diseases using K-means-
based segmentation and neural networks based
classification. Information Technology Journal, 10(2): 267-
275.
[13] Sachin D. Khirade, A. B. Patil, “Plant Disease Detection
Using Image Processing”, 2015 International Conference on
Computing Communication Control and Automation, IEE
[14] Harshal Waghmare,Radha Kokare. “Detection and
Classification of Diseases of Grape Plant Using Opposite
Colour Local Binary Pattern Feature and Machine Learning
for Automated Decision Support System” 2016 3rd
International Conference on Signal Processing and
Integrated Networks (SPIN)
[15]Sumit Nema, Bharat Mishra “Advance App Design
Methods of Leaf Disease Detection using Image Processing
Approach –A Review”International Journal of Innovative
Research in Science,Enginnering and Technologyvol.6,Issue
7,July 2017
[16] Gavhale,K. R.andGawande,U., “An Overview of the
research on plant leaves disease detection using image
processing techniques,” IOSR J. of Computation Engineering
(IOSR-JCE), Vol. 16, pp. 10-16, 2014.

IRJET- Texture based Features Approach for Crop Diseases Classification and Diagnosis-A Research

More Related Content

What's hot (19)

Similar to IRJET- Texture based Features Approach for Crop Diseases Classification and Diagnosis-A Research (20)

More from IRJET Journal (20)

Recently uploaded (20)

IRJET- Texture based Features Approach for Crop Diseases Classification and Diagnosis-A Research