Detect plant diseases using image processing

Other Title(s)

اكتشاف أمراض النباتات باستخدام معالجة الصور

Dissertant

al-Ubaysat, Muna Salih

Thesis advisor

al-Abbadi, Muhammad Ali Husayn

University

Mutah University

Faculty

Information Technology College

Department

Computer Science Department

University Country

Jordan

Degree

Master

Degree Date

2018

English Abstract

Due to the continuous increase in the agricultural area and the evolution of methods, so it is important to study diseases that affect plants and negatively affect the quantity and quality of production of plant crops and field, which is the main source of human food.It is worth noting that the current era of technological developments has led to the exploitation of modern revolutions in the manufacture of modern imaging tools.especially at affordable prices worldwide, as well as the development of scientific research in the so-called classification and computer vision to diagnose diseases with the help of these techniques.

This thesis offers a model to identify the diseases of plant leaves through use of a set of statistical methods to extract discriminatory features.

Our proposed approach depends on the exploitation of many color space models, where the application of four of the most common color channels used: RGB to YCBCR, HSV and HIQ were used to try to identify diseases affecting plant leaves.

In more detail, this model was based on three popular filters; Histogram Equalization, Meanand Median.

After that, the internal parts of the image were then identified and divided into groups after that used the diversity in chromatic properties followed by the application of DWT.

The MATLAB environment was used to implement the proposed model, where “PlantVillage” was benchmark database that used to verify the performance of our approach, this database consisted of (1900) images; (1,300) infected images and (600) healthy images.The performance of the proposed model was evaluated using the confusion matrix, which consists of four components:true negative (TN), true positive (TP), false negative (FN) and false positive (FP), based on this matrix, four of the most popular arbitration and evaluation methods were used:accuracy, F-score, recall and precision.The study found that the proposed model achieved good results in the identification and classification of diseased plants, where the color space models,Ycbcr_HSV_YIQ achieved the best results, it recorded (98.13084112), (96), (100), (98.12265) when Accuracy, Sensitivity, Specificity and F-scorerespectively, while Ycbcr_HSV_YIQ color model achieved the best results where its value was 100% with F measures.

Main Subjects

Civil Engineering

No. of Pages

34

Table of Contents

• APA
• MLA
• AMA

Language

English

Data Type

Arab Theses

Record ID

BIM-1402903