International Journal of Engineering Trends and Applications (IJETA) – Volume 10 Issue 4, Jul-Aug 2023

RESEARCH ARTICLE OPEN ACCESS

Image Processing Techniques in Agriculture for Plant Disease

Detection
Dr. S. Sahaya Tamil Selvi
Associate Professor and Head, Department of computer science
St. Joseph’s college for women- Tirupur- India.
ABSTRACT
Image processing has been proved to be an efficient platform for study in various fields and applications. In the field of
agriculture the quantity and quality of the products are the important factors. The identification of diseases at proper time will
prevent the farmers from loss. This paper discusses about the various methods and techniques involved in the field of image
processing to detect diseases in various types of plants. One of the biggest revolutions of modern history is the invention of
agriculture for a healthier lifestyle. It significantly changed the human culture and played an important role in the development of
the population and biological improvements in food production and domestication.
Keywords - Machine learning, image processing, leaf disease, SVM..

I. INTRODUCTION
preparing is one of the apparatuses which can be connected to
Plant checking is important to control the spread of a malady gauge the parameters identified with agronomy with precision
yet its expense might be high and thus, the makers frequently and economy. Utilizations of picture preparing for farming
skirt basic preventive methodology to keep the creation cost can be comprehensively ordered in two classes: initial one
low. Albeit, official infection acknowledgment is a duty of relies on the imaging systems and the second one dependent
expert agriculturists, minimal effort perception and on applications..
computational helped analysis can adequately help in the
acknowledgment of a plant ailment in its beginning periods. III. LITERATURE ON LEAF BASED
Image processing strategy which is equipped for changing
over a picture into advanced shape and it plays out specific DISEASES DETECTION
tasks on picture, in order to accomplish an upgraded picture or
to extricate some fundamental data from it. The following Srilakshmi [1] [2018], Agriculture assumes an essential job
figure 1. shows the normal and defected leaf. in nation's economy and it has a broad commitment towards
human progress. Because of the developing extensions in
sensor gadgets, RFID and Internet conventions the
engineering of Internet of Things (IoT) has been made to help
farming by making a Smart agribusiness.
Dixit Ekta Gajanan [2] [2018], Although proficient
horticulture engineers are in charge of the acknowledgment of
plant ailments, smart frameworks can be utilized for their
finding in beginning times. The acknowledgment of a sickness
can frequently be founded on manifestations like sores or
spots in different parts of a plant. The shading, territory and
Fig 1. a. Normal Leaf b. Defected Leaf the quantity of these spots can decide, all things considered,
the sickness that has humiliated a plant.
II. IMPLICATION OF IMAGE Saradhambal.G [3] [2018], Crop cultivation plays an
PROCESSING IN AGRICULTURE essential role in the agricultural field. Presently, the loss of
food is mainly due to infected crops, which reflexively
Accuracy agribusiness is another and creating innovation
reduces the production rate.We propose an enhanced k-mean
which prompts joining the propelled systems to upgrade to
clustering algorithm to predict the infected area of the leaves.
cultivate yield and furthermore enhance the homestead
A color based segmentation model is defined to segment the
contributions to productive and naturally sensible way. Ranch
infected region and placing it to its relevant classes.
inputs were imperative parameters to be controlled and if not
Experimental analyses were done on samples images in terms
will result in unfriendly impacts causing decrease in yield,
of time complexity and the area of infected region. Plant
crumbling plant well-being, and so forth. Water system/Water
diseases can be detected by image processing technique.
pressure, Fertilizers, pesticides and nature of yield were the
central point of worry in horticulture. More often than not the
aptitude is required to break with the issues, which might be
tedious and exorbitant issue in creating nations. Picture

ISSN: 2393-9516 www.ijetajournal.org Page 1

 International Journal of Engineering Trends and Applications (IJETA) – Volume 10 Issue 4, Jul-Aug 2023
IV. PROPOSED METHODOLOGY c. Image Segmentation
Plant illnesses can expand the expense of agrarian In computer vision, image segmentation is the process of
generation and may reach out to add up to financial debacle of partitioning a digital image into multiple segments. The goal
a maker if not restored properly at beginning times. The of segmentation is to simplify and change the representation
makers need to screen their yields and identify the primary of an image into something that is more meaningful and easier
indications with the end goal to keep the spread of a plant to analyze. Image segmentation is typically used to locate
illness, with minimal effort and spare the significant piece of objects and boundaries in image. In this work segmentation
the creation. In this proposed framework we have given an algorithm has been adopted.
answer to recouping from the leaf illnesses and furthermore
demonstrate the influenced piece of the leaf by picture
preparing method.

Fig. 3 Capturing leafs are stored in gallery

d. Feature Extraction
The segmented image is first selected by manual
interference. The affected area of the image can be found from
calculating the area connecting the components. First, the
connected components with 6 neighborhood pixels are found.
Later the basic region properties of the input binary image are
found. The interest here is only with the area. The affected
area is found out. The percent area covered in this segment
says about the quality of the result.

e. Classification
Fig 2. Block diagram for leaf defect detecction The current framework can just recognize the sort of
a. Image acquisition ailments which influence the leaf. The proposed leaf infection
The first stage of any vision system is the image acquisition characterization is utilized by Support Vector Machine (SVM)
stage. After that sample images are obtained or collected from classifier calculation. By the utilization of Support Vector
the farm of grape using different mobile cameras with Machine (SVM) classifier calculation, the tainted locale of the
different resolutions. Which are used to train the system. leaf is divided and dissected. The pictures are sustained to this
These sample images are stored in standard jpg Joint application for the grouping of the maladies. It gives a decent
Photographic Experts Group (jpg) format. All the sample decision on horticulture network especially in remote towns. It
images are in RGB (Red, Green, and Blue) form. Collected goes about as a proficient framework as far as arranging the
images include the healthy leaf as well as affected leaf by tainted district. Highlight grouping system serves to
different diseases like bacterial spot, black measles, black rot, classifications the tainted leaf and furthermore to order the
leaf blight, leaf curl virus etc. Various methods of processing plant infections. These are altogether finished with the
can be applied to the image to perform the many different assistance of the advanced mobile phone to catch the pictures.
vision tasks required.
b. Image Pre-Processing Support Vector Machines (SVMs), unmistakably a standout
An input image has some unwanted noise as well as amongst the most mainstream and compelling machine
redundancy present in it. So pre-processing techniques are learning calculations broadly utilized in grouping and
used for noise removal, contrast enhancement and acknowledgment errands in administered learning. The SVM
illumination equalization. To remove the background noise as calculation predicts the event of coronary illness by plotting
well as to suppress the undesired distortion this is present in it. the ailment, foreseeing properties in multidimensional hyper-
These types of variations are occurred due to many reasons plane and characterizes the classes ideally by making the edge
such as camera settings, variation in light etc. of two information groups. This calculation achieves high
exactness by the use of nonlinear capacities called parts. The

ISSN: 2393-9516 www.ijetajournal.org Page 2

 International Journal of Engineering Trends and Applications (IJETA) – Volume 10 Issue 4, Jul-Aug 2023
technique utilizes many shading portrayals all through its Metric Formulas
execution. The division of leaves and foundation is performed
by a Multilayer Perceptron (MLP) neural system that is Accuracy (TP+TN)/(TP+TN+FP+FN)
including a shading library structured from the earlier by
proposing that of an unsupervised Self-sorting Out Guide Sensitivity TP/(TP+FN)
(SOM). Specificity TN/(TN+FP)
V. RESULT AND DISCUSSIONS Precision TP/(TP+FP)

Based on the evaluation metric, the metrics are calculated and the
Recall TP/(TP+FN)
success rate is measured. Classification. Fmeasure 2*Precision*Recall/Precision+Recal
Gmean √Sensitivity+Specificity
Table 1: Evaluation Metrics

Parameters Existing Proposed

Accuracy 65.0631 80.3550

Sensitivity 65 88.4443

Specificity 57.8767 76.8687

Precision 20.6471 39.7543

Fig. 4 Sample classification output for normal leaf
Recall 65 88.4443

Fmeasure 38.0860 65

Gmean 63.3535 74.9074

Table 2: Performance Evaluation Metrics of Normal leaf.

Parameters Existing Proposed
Accuracy 64.0541 85.9780

Sensitivity 60 80.3233
Fig. 5 Sample classification output for defected leaf
Specificity 60.8767 75.9667

Precision 27.6871 45.7453

Recall 60 80.3233

Fmeasure 38.0870 60

ISSN: 2393-9516 www.ijetajournal.org Page 3

 International Journal of Engineering Trends and Applications (IJETA) – Volume 10 Issue 4, Jul-Aug 2023
Gmean 55.3835 77.8024 the plant. The proposed gives the detection with the accuracy
of 85.9570.

REFERENCES
Table 3: Performance Evaluation Metrics of diseased leaf.
[1] Rothe, P. R., and R. V. Kshirsagar. "Adaptive neuro-fuzzy
inference system for recognition of cotton leaf diseases." 2014
Innovative Applications of Computational Intelligence on
Power, Energy and Controls with their impact on Humanity
(CIPECH). IEEE, 2014.
[2] Jafari, M., S. Minaei, and N. Safaie. "Detection of pre-
symptomatic rose powdery-mildew and gray-mold diseases
based on thermal vision." Infrared Physics & Technology 85
(2017): 170-183.
[3] Saradhambal.G, “Plant Disease Detection And Its
Solution Using Image Classification”,IJPAM, Volume 119
No. 14 2018, 879-884 Ferentinos, Konstantinos P. "Deep
learning models for plant disease detection and diagnosis."
Computers and Electronics in Agriculture 145 (2018): 311-
318.
[5] Lv, Jidong, and Liming Xu. "Method to acquire regions
of fruit, branch and leaf from image of red apple in orchard."
Modern Physics Letters B 31.19-21 (2017): 1740039.
[6].Narendra V G, Hareesh K S, “Prospects of computer
Fig. 6. Performance Evaluation Metrics of Normal leaf. vision automated grading and sorting systems in agricultural
and food products for quality evaluation”, International
Journal of Computer Applications,Volume 1 – No. 4, pp 1-12,
2010.
[7]. Mutlu Ozdogan , Yang Yang, George Allez and Chelsea
Cervantes, “Remote Sensing of Irrigated Agriculture:
Opportunities and Challenges”, Remote Sensing, 2, pp- 2274-
2304, 2010.
[8]. M. Omid, M. Abbasgolipour, A. Keyhani and S.S.
Mohtasebi, “Implementation of an Efficient Image Processing
Algorithm for Grading Raisins”, International Journal of
Signal and Image Processing, Vol 1, Issue 1, pp- 31-34, 2010.

Fig. 7. Performance Evaluation Metrics of diseased leaf

IV. CONCLUSION
The work uses to investigate the leaf infection
arrangement at an inopportune environment. The SVM
characterization calculation has been proposed to order the
malady in the plant disease detection system. Plant ailments
can be identified and arranged by image handling strategy by
utilizing the element of the ailment. This framework works
well for the detection of infection which was influenced in

ISSN: 2393-9516 www.ijetajournal.org Page 4