Plagiarism Checker X Originality

Report
Similarity Found: 38%

Date: Wednesday, January 10, 2018

Statistics: 998 words Plagiarized / 2620 Total words
Remarks: Medium Plagiarism Detected - Your Document needs Selective
Improvement.
------------------------------------------------------------------------------------------
-

Leaf Disease Detection Based On Machine Learning Anish Polke Kavita Joshi Dr.
Pramod Gouda Dept. of E&TC Dept. of E&TC Dept. of E&TC GHRCEM, Pune, India
GHRCEM, Pune, India GHRCEM, Pune, India
polke_anish.ghrcemmevlsi@raisoni.net kavita.joshi@raisoni.net
pramod.gouda@raisoni.net Abstract — Agriculture is one of the most important
factors bas nwacontryecomyis de nnts n crops are very regular, which is the one
of the prominent factor that leads to the disease location and detection in plants
pats is of high importance in agroindustry.

Therefore it is important to successfully identify the diseases from the crop to

selectively spray herbicides and fertilizes to reduce wastage use of chemical. In
this work, we present an approach that integrates image processing and machine
learning to allow diagnosing diseases from leaf images, to study quantitative
plant physiology imaging and computer vision is used. Wavelet is very popular
tool in image processing algorithm. Texture features are used for detection of
crop.

These features are mean, standard deviation, skewness and kurtosis which we
have used in this paper. We are proposing an approach, which used to identify
the plant infection i.e. plant disease. Here we are using minimum distance
classifier for the classification of the disease.
The proposed approach presents a path toward automated plant diseases
diagnosis on a massive scale. Keywords — image segmentation, plant disease,
minimum distance classifier. I. INTRODUCTION Indian economy is highly
dependent of agricultural productivity. One of the major factors responsible for
the crop destruction is plant disease.

Present method for disease detection of plant is observation by experts through

their naked eye which needed; a large team of experts as well as continuous
monitoring of plant is required, which is costly affair. In developed countries they
are using the cutting edge technologies like information and space technology,
biotechnology. It is necessary to change and utilize the research for eco-friendly
crop management and increased role of research-based technologies and
sustainable crop production.

In plants, some general diseases seen are brown and yellow spots, early and late
scorch, and others are fungal, viral and bacterial diseases. Image processing [1]
can be used for measuring affected area of disease and to determine the
difference in the color of the affected area. Automatic detection of the diseases
by just seeing the symptoms on the plant leaves makes it easier as well as
cheaper.

Therefore in field of agriculture, use of automatic disease detection technique in

plants plays an important role. Detection of a plant disease in very initial stage
would be very beneficial. This also supports machine vision to provide image
based automatic process control, inspection, and robot guidance.

The simple principle of the operation of proposed system is to take the responses
from various parameters that decide the productivity, processing them according
to the algorithm and predict the suitable crop for the land along with the current
status of crop if infected or not. Also suggestion is given of some fertilizers that
could be used to improve the fertility.

For making system available to common people we propose this system is made
as an Android Application, where the farmer could feed the inputs, and obtain
the necessary result. II. LITERATURE SURVEY Literature review reports that the
machine vision research varies widely in depending on the complexity of the
visual scene. With the advancement of the personal computer most of the on
field machine vision detection work has been performed and resulted that is
starting from individual leaves kept on plain background to narrow leaves scenes
of field plants growing in on field natural conditions.
Textures and size feature used for weed detection and automation of fertilizer
spraying [2]. Wavelet transform is used to extract the texture features of the crop
and weed images for classification [3]. Energy, Entropy, Inertia, Contrast,
Homogeneity these five texture features are extracted. Vijai Singh and A.K.

Misra [4] presents the review on various diseases classification techniques utilized
for plant leaf illness location and a calculation for picture division strategy that
can be utilized for programmed identification and additionally characterization of
plant leaf ailments later. Banana, beans, jackfruit, lemon, mango, potato and
tomato, are some of those ten species on which proposed calculation is tried.

Along these lines, related ailments for these plants were taken for ID. With less
computational endeavors the ideal outcomes were acquired, which likewise
demonstrates the proficiency of proposed calculation in acknowledgment and
characterization of the leaf illnesses.

Another favorable position of utilizing this technique is that the plants illnesses
can be distinguished at beginning period or the underlying organize. To enhance
acknowledgment rate in grouping process Simulated Neural Network, Bayes
classifier, Fuzzy Logic and cross breed calculations can likewise be utilized. S.Raj
Kumar and S.Sowrirajan [5] have proposed image-processing based way to deal
with consequently order the ordinary or infected leaves (Early leaf spot, Late leaf
spot, Alternaria leaf spot) and furthermore give the cure to a similar which would
be gainful to novices in cultivating or planting (as these maladies are basic in
blossoming plants like rose too) .

In they approach they have joined all the mixture highlights of a leaf to prepare
the ANN (BPN-FF) and have made utilization of Lloyd's bunching which is more
productive than the customary K-implies grouping to section the test pictures.
Dheeb Al Bashish et. al. [6], proposed that the RGB images be converted into HSI
plane and then the color features are extracted (by SGDM generation).

The texture features are extracted by obtaining GLCM (Grey Level Co- occurrence
Matrix). The input images are segmented using K-means clustering technique
and then the segmented images are analyzed by a pre-trained BPN network for
detection and classification of plant leaf and stem diseases like early scorch,
cottony mold, ashen mold, late scorch and tiny whiteness.

The author also compares between various models incorporating various

components such as HS,H,S,I,HSI and found that model HS provides the best
efficient output amongst all other models with efficiency of 92.7%. Niket, A.,et. al.
[7], stated that the RGB images upon acquisition undergo color space
transformation into HSI plane and upon segmentation using K-means clustering
the green pixels are masked from the appropriate cluster and the masked green
pixels are removed. Then the useful segments are obtained.

Then the texture features are extracted using Color-Co-Occurrence Matrix (CCM).
The classifier used in BPN-FF. But addition color or shape features or both might
have improved the efficiency of classification. Arivazhagan, S.,et. al. [8], proposed
a new method in which the RGB images are converted to HSI plane and only the
Hue component is used for further analysis.

Then the green pixels are masked and the masked green pixels are removed. The
useful components are obtained upon segmentation and only the texture feature
is extracted using Co-occurrence matrix. Then the neural network employing SVM
classifier is used to detect and classify early scorch, yellow spots, brown spots,
late scorch with efficiency of 94.74%.

Here only a single feature extraction is employed and the classifier which is not
that efficient in classifying the disease but effectively detects whether the leaf is
diseased or not. Gurpreet Kaur and Himanshu, M.,et. al. [9], provide the study of
various classifiers namely K-Nearest Neighbors(KNN) classifier, Support Vector
Machines(SVM) classifier, Back Propagation Neural Network-Feed Forward(BPN-
FF) classifier, Probabilistic Neural Network(PNN) classifier, General Regression
Neural Network(GRNN) classifier. Sanjeev, S.S.,et. al.

[10], provided a new way in which the k-means segmentation is done, followed
by feature extraction using GLCM and the classification is done via BPN. Here
only hue component is used and instead of k- means segmentation other
algorithms could have employed to extract lesion more accurately. G.
Ravichandran and Koteeshwari R S [11], built android app for the crop prediction
which suggest the suitable crop to farmer based on soil parameters, which were
entered in application by farmer after soil testing. Here automatic soil testing is
not performed and there is no check on the current crop status. III.

PROPOSED SYSTEM We are proposing a system which is a combination of

hardware and software. For obtaining the physical attributes we use sensors like
pH and soil moisture sensors, which are interfaced with controller. Here we are
using raspberry pi controller. Figure 1 shows the block diagram of proposed
system. Figure 1.

Block Diagram Of The System Digital Camera is used to capture the images from
farm and transfer them to computer where image processing is performed in
order to detect any diseases on the plant leaves. If starting of disease is detected
then appropriate fertilizer is sprinkled on crop using fertigation pump. As we are
having pH and moisture related data of soil, suggestion can be given to farmer
which crop is suitable of that soil.

We have gathered the dataset of over 200 high resolution images of leaves. First
preprocessing of images is performed to remove any noise in image to perform
enhancement of image, then we perform segmentation using the conversion of
image from RGB to YCBCR plane and feature extraction of where we have
considered features such as mean, standard deviation, kurtosis and skewness for
texture features and discrete wave form transformation for the colour feature
extraction.

Classifier used is minimum distance classifier which is used to classify unknown

image data to classes which minimize the distance between the image data and
the class in multi-feature space. The distance is defined as an index of similarity
so that the minimum distance is identical to the maximum similarity with the
reference image.

Based on which decision is taken if leaf is having any disease or in healthy state.
IV. RESULT ANALYSIS The result of the proposed system is given below. After
training the dataset the pre-processing of the input image is done by using the
top hat and bottom hat approach the below figure 2 shows the top hat and
bottom hat enhancement results.

The enhancement is used for enhancing the features of the raw image after the
enhancement the features of the input image is get enhanced and the level of
noise is get reduces [12]. Figure 2. Output of final Enhancement After the
enhancement the image segmentation is performed which is used for
segmentation the image various contours after the segmentation the features are
extracted. The output of YCbCr model is given in figure 3.

Y = (77/256)R + (150/256)G + (29/256)B -(1) Cb = -(44/256)R - (87/256)G +

(131/256)B + 128 -(2) Cr = (131/256)R - (110/256)G - (21/256)B + 128 -(3) Figure
3. Output of YCbCr model We have calculated the features of image matrix for
example. Mean which is for a random variable vector A made up of N scalar
observations, the mean is defined as [13].

The Standard Deviation is a measure of how spreads out numbers are. Skewness
is a measure of the asymmetry of the data around the sample mean. If skewness
is negative, the data are spread out more to the left of the mean than to the
right. If skewness is positive, the data are spread out more to the right. The
skewness of the normal distribution (or any perfectly symmetric distribution) is
zero.

The skewness of a distribution is defined as Kurtosis is a measure of how outlier-

prone a distribution is. The kurtosis of the normal distribution is 3. Distributions
that are more outlier-prone than the normal distribution have kurtosis greater
than 3, distributions that are less outlier- prone has kurtosis less than 3. The
kurtosis of a distribution is defined as Figure 4.

RGB Histogram After the segmentation the feature extraction is done, here we
are calculating the color feature and color moments using the RGB to HSV
conversion the output of the colour feature extraction is shown in figure 5. The Cr
channel image is multiplied with complete back image for masking and we can
see only diseased part of leaf, however this image is without colour, therefore we
again multiply it with enhanced image pixel by pixel and we got below colour
segmented image. Figure 5.

Colour segmented image After color feature extraction, we move on to extract

the texture feature. For this purpose we make use of the discrete wavelet
transform. The discrete wavelet transforms states to wavelet transforms that the
wavelets are disjointedly appraised. A transform which limits a function both in
space and scaling and has some necessary properties compared to the Fourier
transform.

The transform is centered on a wavelet matrix, which can be figured more quickly
than the analogous Fourier matrix. Most notably, the DWT [14] is used for signal
coding, where the assets of the transform are exploited to signify a discrete signal
in an extra redundant form, often as a preconditioning for data compression. The
discrete wavelet transform has a vast quantity of applications in science,
computer science, mathematics and engineering.

The mean and standard deviation of the distribution of the wavelet transform
coefficients are used to construct the feature vector. . Figure 6. Output of Feature
Extraction V. CONCLUSION To sustain high productive growth in a sustainable
manner in agriculture, there is a need to move from traditional intensive to
technology and skill intensive agriculture techniques, with the advancement in
the digital camera and smartphones, we can use leverage there images in image
processing.

The process of detection and diagnose of disease from such image has been
active research area useful in development of several applications to help farmers
to know about diseases affecting their yield and save cost needed for the expert
suggestions. Despite of several techniques available for disease diagnose of
plants, more scope of the project lies to develop computationally inexpensive,
robust and high detection and recognition rate techniques. VI. REFERRENCES [1]
ViutA n odhSK. Apliaonof mapessig agriculture: a survey ” , International Journal
of Computer Applications.

52(2), pp. 34-40, 2012. [2] Amruta A. Aware adKat i“ Crop and Weed Detection
Based on Texture and Size Features and Automatic Spraying of Herbicides ”,
International Journal of Advanced Research in Computer Science and Software
Engineering, Volume 6, Issue 1, January 2016. [3] Amruta A.

Aware and Kavita Joshi “ Wavelet Based Crop Detection And Automatic Spraying
Of Herbicides ” , International Journal of Innovations & Advancement in
Computer Science, Volume 4, Issue2, February 2015. [4] Vijai Singh and A.K. Misra
“ Detection of plant leaf diseases using image segmentation and soft computing
techniques Information Processing in Agriculture 4 (2017) 41 – 49. [5] S.Raj Kumar
and S.raan“Auomat f seaDett ion a lafiti sinHybdFeaua ervi lafi International
Journal of Advanced Research in Electrical, Electronics and Instrumentation
Engineering Vol.

5, Issue 6, June 2016 DOI:10.15662/IJAREEIE.2016.0506008. [6] Dheeb Al Bashish,

M akBikanSuemanB. Fra Dettona lafiatonof nt f nStDises”, IEEE International
Conference on Signal and Image Processing, 2010. [7] Niet
,BhratJadhv,Smeet,N.“Dettona lafiti plant dis eabipessig”, International Journal of
Innovative Science, Engineering & Technology, Vol. 1 Issue 2, April 2014. [8]
Arivazhagan, S.,

NewlinShebiah, R., Ananthi, S., Vishnu Varthini,S., “Dettonof eahregi pntleaa lafic
ation of plant leaf diseases using tufeau CIGR Journal, Vol.15, March 2013. [9]
Gureetu man,M. lafiti bologiaspi bsedonleaahecu, International Journal of
Computer Science and Information Technology & Security (IJCSITS), Vol. 2, No.2,
April 2012. [10] Saj ,Viy,S. rgud,V.B,Pavi“Digns cssiaonof plf ises sig eul et” IEEE-
2013 4th ICCCNT, July 4-6, 2013.

[11] Giritharan Ravichandran and Koteeshwari R S “ Agricultural Crop Predictor

and Advisor using ANN for Smartphones ” , IEEE International Conference on
Emerging Trends in Engineering, Technology and Science (ICETETS), February
2016. [12] Malik, S., & Lone, T., “ Comparative study of digital image
enhancement approaches ” , International Conference on Computer
Communication and Informatics 2014.
http://dx.doi.org/10.1109/iccci.2014.6921749.

[13] Mrc ti,MhBaud grdDaeches Image odn ug ve rasform , IEEE transaction on

Image Processing, Volume 1 No. 2 April 1992. [14] Sajad Kiani1, Zohreh Azimifar
and Saadat Kamgar, “ Wavelet based crop detection and classification ” , ICEE
2010, May 11-13, IEEE 2010.

INTERNET SOURCES:
------------------------------------------------------------------------------------------
-
0% - Empty
0% - http://www.science.gov/topicpages/s/sign
0% - https://www.britannica.com/science/disea
0% - https://broom02.revolvy.com/topic/Herbic
1% - http://ieeexplore.ieee.org/document/7946
0% - https://www.cs.cmu.edu/~cil/v-source.htm
0% - https://www.a-research-paper.com/paper/d
0% - http://www.apsnet.org/edcenter/intropp/t
1% - http://www.ijcta.com/documents/volumes/v
1% - http://ieeexplore.ieee.org/document/7946
5% - https://www.sciencedirect.com/science/ar
0% - https://aggie-horticulture.tamu.edu/vege
5% - https://www.sciencedirect.com/science/ar
0% - https://www.indiamart.com/viezec/service
0% - https://www.researchgate.net/post/How_ca
5% - https://www.sciencedirect.com/science/ar
5% - https://www.sciencedirect.com/science/ar
5% - https://www.sciencedirect.com/science/ar
5% - https://www.sciencedirect.com/science/ar
5% - https://www.sciencedirect.com/science/ar
0% - https://www.ctahr.hawaii.edu/oc/freepubs
0% - http://www.cs.ox.ac.uk/teaching/studentp
0% - https://www.sciencedirect.com/science/ar
0% - https://www.scribd.com/document/34437168
2% - http://www.ijritcc.org/download/browse/V
0% - https://www.scribd.com/document/35571679
0% - https://www.ijeas.org/download_data/IJEA
5% - https://www.sciencedirect.com/science/ar
0% - http://www.plantphysiol.org/content/148/
0% - https://www.scribd.com/document/33645075
0% - http://www.interfire.org/res_file/lab.as
0% - https://www.sciencedirect.com/science/ar
0% - https://www.facebook.com/rajkumar.rajkum
0% - http://www.academia.edu/7671462/An_Autom
0% - http://www.irdindia.in/journal_ijraet/pd
1% - https://www.scribd.com/document/87930055
0% - http://www.ijera.com/papers/vol%201%20is
0% - http://www.essay.uk.com/essays/environme
1% - https://www.scribd.com/document/87930055
0% - http://ascelibrary.org/doi/10.1061/%28AS
0% - http://www.ijsrd.com/articles/IJSRDV2I93
0% - http://ijarcet.org/wp-content/uploads/IJ
0% - http://www.ijcttjournal.org/Volume15/num
0% - https://ijirset.com/upload/2014/june/104
0% - http://www.cell.com/trends/cognitive-sci
0% - https://jp.mathworks.com/matlabcentral/a
5% - https://www.sciencedirect.com/science/ar
0% - http://www.cisjournal.org/journalofcompu
0% - https://www.sciencedirect.com/science/ar
0% - https://www.ijser.org/paper/Fuzzy-Logic-
0% - http://www.e-mjm.org/2014/supplement-A/m
0% - http://research.ijcaonline.org/volume111
0% - https://www.science.gov/topicpages/b/bac
0% - http://www.enggjournals.com/ijet/docs/IJ
0% - https://link.springer.com/article/10.100
0% - http://www.cwejournal.org/category/vol9n
0% - http://agritech.tnau.ac.in/org_farm/orgf
0% - http://dspace.cusat.ac.in/jspui/bitstrea
0% - https://in.pinterest.com/opencrm/technol
0% - https://www.coursehero.com/file/p61g4t/F
0% - https://www.google.com/patents/US7222357
1% - http://www.ee.columbia.edu/~cylin/course
0% - https://www.scribd.com/document/29637593
0% - https://link.springer.com/chapter/10.100
1% - http://www.ijcta.com/documents/volumes/v
1% - http://esatjournals.net/ijret/2015v04/i0
1% - http://cisjournal.org/journalofcomputing
0% - http://www.uptodate.com/contents/heart-t
0% - http://www.scielo.br/pdf/babt/v59nspe2/1
0% - http://shodhganga.inflibnet.ac.in/bitstr
0% - https://ijirset.com/upload/december/2-Re
0% - http://www.ijert.org/download/8717/facia
0% - http://web.stanford.edu/~wfsharpe/mia/ma
1% - https://www.mathworks.com/help/matlab/re
1% - https://www.mathworks.com/help/stats/ske
1% - https://gist.github.com/hillar/17520201b
1% - https://es.mathworks.com/help/stats/skew
2% - https://www.mathworks.com/help/stats/kur
2% - https://www.mathworks.com/help/stats/kur
2% - https://www.mathworks.com/help/stats/kur
0% - http://www.rroij.com/open-access/image-r
1% - https://www.irjet.net/archives/V2/i1/Irj
0% - https://www.scribd.com/document/35102296
0% - http://www.mif.vu.lt/atpazinimas/dip/FIP
0% - https://www.researchgate.net/profile/Rin
2% - http://www.ijcta.com/documents/volumes/v
1% - http://www.jatit.org/volumes/Vol68No1/25
1% - http://www.jatit.org/volumes/Vol68No1/25
2% - http://www.ijcta.com/documents/volumes/v
2% - http://www.ijcta.com/documents/volumes/v
1% - http://ijarse.com/images/fullpdf/1512906
0% - http://citeseerx.ist.psu.edu/viewdoc/dow
0% - https://quizlet.com/36752216/u5-aphg-fla
0% - http://iosrjournals.org/iosr-jece/papers
0% - http://www.ijcmas.com/vol-3-10/Nikunj%20
0% - http://adsabs.harvard.edu/abs/2012IJCA..
2% - http://www.ijritcc.org/download/browse/V
1% - https://www.researchgate.net/profile/Rah
2% - http://www.ijritcc.org/download/browse/V
0% - http://academicscience.co.in/admin/resou
0% - https://www.springerprofessional.de/en/w
0% - https://www.researchgate.net/post/what_i
0% - http://www.rroij.com/indexing-internatio
0% - https://arxiv.org/pdf/1001.2267
1% - http://toc.proceedings.com/31976webtoc.p
0% - http://www.academic.net/show-10-1310-1.h
0% - https://www.researchgate.net/publication
0% - http://files.matlabsite.com/docs/papers/