International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056

Volume: 05 Issue: 04 | Apr-2018 www.irjet.net p-ISSN: 2395-0072

Detection and Analysis of Plant Diseases Using Image Processing

Technique
Shweta R. Astonkar1, Dr. V. K. Shandilya2

Department of computer science and engineering, Sipna COET, Maharashtra, India

---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - Identification of the plant diseases is the key to Agriculture in India is the means of livelihood of almost two
preventing the losses in the yield and quantity of the thirds of the workforce in India. India has over 210 million
agricultural product. The studies of the plant diseases mean acres of farm land. Jawar, wheat, sunflower, cereals are the
the studies of visually observable patterns seen on the plant. major crops. Apple, banana, sapota, grapes, oranges are the
Health monitoring and disease detection on plant is very most common fruits. Sugarcane, cotton, chili, groundnuts are
critical for sustainable agriculture. It is very difficult to the major commercial crops. Computer Vision Systems (CVS)
monitor the plant diseases manually. It requires tremendous developed for agricultural applications, namely, detection of
amount of work, expertize in the plant diseases, and also weeds, sorting of fruits in fruit processing, classification of
require the excessive processing time. Hence, image processing grains, recognition of food products in food processing,
is used for the detection of plant diseases. Disease detection medicinal plant recognition, etc. In all these techniques,
involves the steps like image segmentation, feature extraction digital images are acquired in a given domain using digital
and classification. This paper discussed the method used for camera and image processing techniques are applied on
the detection of plant diseases using their leaves images. these images to extract useful features that are necessary for
further analysis. Plant disease diagnosis is an art as well as
Agriculture is a most important and ancient occupation in science. Many diseases produce symptoms, which are the
India. As economy of India is based on agricultural production, main indicators in field diagnosis. The diagnostic process
utmost care of food production is necessary. Pests like virus, (i.e., recognition of symptoms and signs), is inherently visual
fungus and bacteria causes infection to plants with loss in and requires intuitive judgment as well as the use of
quality and quantity production. There is large amount of loss scientific methods. The photographic images of symptoms
of farmer in production. Hence proper care of plants is and signs of plant’s diseases used extensively to enhance
necessary for same. description of plant diseases are invaluable in research,
teaching, diagnostics, etc. Plant pathologists incorporate
digital images using digital image transfer tools in diagnosis
Key Words: Image processing, Plant Diseases, Agriculture
of plant diseases. Till now experts identify the presence of
the disease in the plants manually, but it is expensive for a
1. INTRODUCTION farmer to consult an expert due to their distant availability,
so it is required to detect the symptoms of the plant diseases
India is a cultivated country and about 70% of the
automatically as early as they appear on the plant. Early
population depends on agriculture. Farmers have large detection will help farmers to avoid huge loss. Technology
range of diversity for selecting various suitable crops and support would help them in early detection of diseases,
finding the suitable pesticides for plant. Disease on plant cutting on cost of pesticides, and good returns for the efforts,
leads to the significant reduction in both the quality and thus making the profession attractive. To remedy this
quantity of agricultural products. The studies of plant situation various alternatives are being searched to
disease refer to the studies of visually observable patterns minimize the application of these hazardous chemicals. One
on the plants. Monitoring of health and disease on plant of the main concerns of scientists is the automatic disease
plays an important role in successful cultivation of crops in diagnosis and control. Several key technologies
the farm. In early days, the expertise person in that field did incorporating concepts from image processing and artificial
the monitoring and analysis of plant diseases manually. This intelligence are developed by various researchers in the past
requires tremendous amount of work and requires excessive to tackle this situation.
processing time. The image processing techniques can be
used in the plant disease detection. In most of the cases
disease symptoms are seen on the leaves, stem and fruit. The We can analyze the leaf diseases by using hardware
plant leaf for the detection of disease is considered which techniques like agriculture robot which will capture the
shows the disease symptoms. This paper gives the images of crop by using webcam and with the help of those
introduction to image processing technique used for plant images the robot built model detect the appropriate
disease detection. diseases. However, there exist some drawbacks which will
not be able to give appropriate result. The webcam is not
Today India ranks second worldwide in farm output. able to capture the accurate and clear image because of
Agriculture is still the largest economic sector and plays a pixels. But accurate detection of diseases is important.
major role in socioeconomic development of India. Rather than using webcam in hardware model we can

© 2018, IRJET | Impact Factor value: 6.171 | ISO 9001:2008 Certified Journal | Page 3190
 International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 05 Issue: 04 | Apr-2018 www.irjet.net p-ISSN: 2395-0072

propose a software which is based on neural network. In this acknowledgment is accomplished through PCA or GLCM and
project we are going to develop a system which detect the arrangement through SVM or ANN. Procedures, for example,
diseases on leaf in which the images will be capture by smart middle channel, histogram balance, picture smoothening,
phones because in today’s era everyone has smart phones image honing, and so on., can be utilized for performing
hence every farmer will be able to capture images and picture improvement. A portion of the calculation
upload those to detect diseases. This software will solve and procedures that can be utilized for surface element
answer different type queries and particular person can ask extraction are Gabor channel, shading co-event techniques,
queries in their own language for his/her convenience. If wavelet change, and so forth.
they are looking for the remedies of particular disease the
they can login and get the remedies. 3. STEPS FOR PLANT DISEASE DETECTION
2. LITERATURE SURVEY In this section, the basic steps for the disease detection as
shown in fig:
Monica Jhuria et al uses image processing for
detection of disease and the fruit grading in [1]. They have
used artificial neural network for detection of disease. They
have created two separate databases, one for the training of
already stored disease images and other for the
implementation of the query images. Back propagation is
used for the weight adjustment of training databases. They
consider three feature vectors, namely, color, textures and
morphology [1]. They have found that the morphological
feature gives better result than the other two features.

Thus, study of various plant diseases means

studying visual pattern because it is difficult to monitor
diseases on plant manually. Basic steps like image
acquisition, preprocessing, segmentation, extraction;
detection and classification of plant disease are used [2]. K-
means clustering followed by thresholding and feature
extraction is described. Classification by using neural
network ANN and BPNN has also been used for detection of
disease

For proper and successful cultivation of crops it is

necessary to detect diseases accurately [3]. Hence, from
above discussion it can be seen that image-processing
techniques have proved useful in all means. We can
accurately detect and classify diseases on various plants
using all above techniques. K-means Clustering to detect Fig 1: Steps for Plant Disease Detection
infected objects and Neural Networks are thus commonly
used for obtaining accuracy in detecting and classifying [A] Image Preprocessing
diseases. Hence, these techniques have potential to be use in
Agrobot system. Image preprocessing is to remove noise from the image or
other object removal, different preprocessing techniques.
Di Cui et al. [4], explains the image processing Here we are using image scaling. Image scaling is used to
techniques for multispectral images to detecting rust on convert the original image into thumbnails because the pixel
plant leaf and its growth amount of disease. The dataset size of original image is large and it require more time for
contains the images collected from a greenhouse of research the overall process hence after converting the image into
institute. The explained method uses the concept of evaluate thumbnails the pixel size will get decreases and it will
centroid for each image for further processing. require less time.

[B] Image Segmentation

Rittika Raichaudhuri et al. [5] this venture is to give
a framework to identifying wheat leaf Diseases. In this
Image segmentation is one of the mostly used methods to
robotized framework will be utilized to identify the wheat
classify the pixels of an image correctly in a decision-
leaf infections however picture preparing. The k mean
oriented application. It divides an image into a number of
calculation and vigilant channel are utilization for the image
discrete regions such that the pixels have high similarity in
preparing and segmentation handling. Design
each region and high contrast between regions. It is a

© 2018, IRJET | Impact Factor value: 6.171 | ISO 9001:2008 Certified Journal | Page 3191
 International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 05 Issue: 04 | Apr-2018 www.irjet.net p-ISSN: 2395-0072

valuable tool in many field including health care, image 4. arrange Eu [] [] in ascending order
processing, traffic image, pattern recognition etc. There are
4. classify the pixel into the cluster having centroid j for
different techniques for image segmentation like threshold
which the pixel has minimum Euclidian distance
based, edge based, cluster based, and neural network
based1. From the different techniques, one of the most 5. end for
efficient methods is the clustering method. Again, there are
6. calculate centroid mean
different types of clustering: K-means clustering, Fuzzy C-
means clustering, mountain clustering method and 7. check whether any Euclidian distance is greater than
subtractive clustering method. One of most used clustering centroid mean or not.
algorithm is k-means clustering. It is simple and
8. if Euclidian distance of any pixel is greater than centroid
computationally faster than the hierarchical clustering. In
mean then
addition, it can work for large number of variable. However,
it produces different cluster result for different number of 9. make the pixel as centroid and repeat step 4 until the
number of cluster. So it is required to initialize the proper centroids get stable
number of number of cluster, K2. Again, it is required to
10. rearrange clusters to form new-clustered image
initialize the k number of centroid. Different value of initial
centroid would result different cluster. So selection of proper 11. save image on server
initial centroid is also an important task. Nowadays image
segmentation becomes one of important tool in medical area [C] Feature Extraction
where it is used to extract or region of interest from the
background. So medical images are segmented using Feature Extraction is an important part in the disease
different technique and process outputs are used for the detection. It plays an important role in identification of an
further analysis in medical. t plant. object. Feature extraction is used in many applications in
image processing. Color, texture edges, morphology are the
K-Means Clustering: features, which are used in disease detection.

Clustering is a method to divide a set of data into a specific Monica jhuria et al took color, morphology, texture as feature
number of groups. It’s one of the popular method is k-means for the disease detection. It is found that morphological
clustering. In k-means clustering, it partitions a collection of result gives more result than any other features. Texture
data into a k number group of data11,12. It classifies a given shows how the color is distributed in the image, hardness of
set of data into k number of disjoint cluster. K-means the image.
algorithm consists of two separate phases. In the first phase
it calculates the k centroid and in the second phase it takes [D] Detection and Classification of Plant Diseases
each point to the cluster which has nearest centroid from the
respective data point. There are different methods to define The final stage is the detection of the diseases and with the
the distance of the nearest centroid and one of the most used help of disease classify the plants with the disease matches
methods is Euclidean distance. Once the grouping is done it with the given dataset. For the disease detection and
recalculate the new centroid of each cluster and based on classification, we are implementing the deep learning
that centroid, a new Euclidean distance calculated between algorithm. Deep learning algorithm is used to classify the
each center and each data point and assigns the points in the specified image into appropriate disease hence it will be easy
cluster, which have minimum Euclidean distance. Each to detect the disease and find out the remedy over the
cluster in the partition is defined by its member objects and disease. Deep learning algorithm is the part where we are
by its centroid. The centroid for each cluster is the point to finding out the relevancy count of the pixels by comparing
which the sum of distances from all the objects in that the images with the data set. According to the relevancy
cluster is minimized. So K-means is an iterative algorithm in count we will find out the matched disease.
which it minimizes the sum of distances from each object to
its cluster centroid, overall clusters.  Deep Learning Algorithm

K mean Clustering Algorithm 1. upload image and read it into pix [][]
2. Initialize Pix [][] with image pixels
1. Initialize Pix [] with image pixels 3. for i=0 to pix [][]. len
2. select Random Centroids [] a. match pix [][] attributes with dataset
pixels’ attributes
3. For i=0 to pix.len b. store matching result in db
1. for j=0 to Centroids [].len 4. end for
2. calculate Eu[i] [j] = (i, sqrt (Sum (power (pix[i]- 5. calculate weights for every pixel
centroid[j]),2)) 6. calculate z for matching diseases z=sum (w)
3. end for 7. take average of z=avg (z)

© 2018, IRJET | Impact Factor value: 6.171 | ISO 9001:2008 Certified Journal | Page 3192
 International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 05 Issue: 04 | Apr-2018 www.irjet.net p-ISSN: 2395-0072

8. Filter diseases having z value less than avg(z) Proceedings of the International Conference on Data
9. Transfer the output of 1st layer to 2nd layer Engineering and Communication Technology, Advances in
10. take average of z=avg (z) Intelligent Systems and Computing. 2016;1; 978- 981.
11. filter diseases having z value less than avg(z)
12. transfer the output of 1st layer to 3nd layer [6] Wenjiang Huang, Qingsong Guan, Juhua Luo, Jingcheng
13. output of 3rd layer will be delivered to user. Zhang, Jinling Zhao, Dong Liang, Linsheng Huang, and
Dongyan Zhang, “New Optimized Spectral Indices for
4. ADVANTAGES Identifying and Monitoring Winter Wheat Diseases”, IEEE
journal of selected topics in applied earth observation and
1. Efficient and user friendly system. remote sensing, Vol. 7, No. 6, June 2014.

2. Improved accuracy with the help of K mean clustering

Algorithm and Deep learning algorithm.

3. Increased layers of deep learning algorithm to get most

accurate and appropriate result.

4. Our system is very fast as compared to other because

segmentation makes the processing time less.

5. CONCLUSION

We have segmented an image by using k-clustering

algorithm using cluster to generate the initial centroid and
the final segmented result is compare with k-means
clustering algorithm and we can conclude that the proposed
clustering algorithm has better segmentation and after
formation of clusters apply the deep learning algorithm to
find out the matching image, we have found the value of z in
each stage and compare those with the dataset images and
the value having less z value will be the final value. Deep
learning algorithm gives the appropriate result of the
diseases and takes less amount of time for detection than
other methods.

REFERENCES

[1] Monica Jhuria, Ashwani Kumar, and Rushikesh Borse,

“Image Processing for Smart Farming: Detection of Disease
and Fruit Grading”, Proceedings of the 2013 IEEE Second
International Conference on Image Information Processing
(ICIIP-2013.

[2] Sachin D. Khirade, and A. B. Patil, “Plant disease

detection using image processing,” IEEE

[3] Shivani K. Tichkule, Prof. Dhanashri. H. Gawali, “Plant

Diseases Detection Using Image Processing Techniques”,
2016 Online International Conference on Green Engineering
and Technologies (IC-GET).

[4] Di Cui, Qin Zhang, Minzan Li, Glen L. Hartman, Youfu

Zhao, “Image processing methods for quantitatively
detecting soybean rust from multispectral images”,
ELSEVIER-Science Direct, Vol.22, No.4, pp.186193, 2010.

[5] Rittika Raichaudhuri and Rashmi Sharma. "On Analysis

of Wheat Leaf Infection by Using Image Processing"

© 2018, IRJET | Impact Factor value: 6.171 | ISO 9001:2008 Certified Journal | Page 3193