Paddy abnormality vision recognition tool based on multi-layered mamdani fuzzy modeling

Siti Norul Huda Sheikh Abdullah, Raihan Afifah Mohamed Yusoff, Noor Faridatul Ainun Zainal, Saad Abdullah, Azuraliza Abu Bakar, Khairuddin Omar

Research output: Contribution to journalArticle

3 Citations (Scopus)

Abstract

Conflicts often arise between farmers and experts regarding the solution for handling issues on paddy abnormality. Although multiple and serial courses are periodically conducted by the experts, yet the farmers have different ideas when dealing with paddy diseases. The disagreement regarding paddy diseases has made it difficult for Malaysia to meet its goal of producing 20 metric tons/ha. Hence, the objective of this research is to propose an automatic paddy abnormality visual recognition tool. A multi-layered Mamdani fuzzy modeling approach has been used in classifying three common paddy crop diseases, namely Blast, Brown Spot, and Narrow Brown Spot. Healthy leaf class is also included as a controlled parameter. The fuzzy classifier is designed to consist of four fuzzy membership functions by incorporating the characteristics of the lesion from the image gained from the texture analysis. Blob analysis is used in extracting features for the recognition of the lesion shape. Color charts are constructed with the help of an expert based on the RGB color space for the leaf color, lesion spot color, and lesion boundary color. In assessing a lesion, the best color match to the color charts is determined by using the Euclidean distance. An experiment is conducted where the performance of the proposed approach is compared to other classifiers which are ID3, Multi-layer Perceptron, Naïve-Bayes and Decision Table/Naïve-Bayes. This article proves that the proposed approach has the highest accuracy rate when compared to the other classifiers which is 85.80%, with 87.01% sensitivity and 81.24% specificity rate, respectively.

Original languageEnglish
Pages (from-to)325-334
Number of pages10
JournalApplied Engineering in Agriculture
Volume30
Issue number2
DOIs
Publication statusPublished - 2014

Fingerprint

Color
Classifiers
Decision tables
Multilayer neural networks
Membership functions
Crops
Textures
Experiments

Keywords

  • Crop disease
  • Inspection
  • Multi-layered mamdani fuzzy modeling
  • Pattern recognition
  • Texture analysis

ASJC Scopus subject areas

  • Engineering(all)

Cite this

Paddy abnormality vision recognition tool based on multi-layered mamdani fuzzy modeling. / Sheikh Abdullah, Siti Norul Huda; Yusoff, Raihan Afifah Mohamed; Zainal, Noor Faridatul Ainun; Abdullah, Saad; Abu Bakar, Azuraliza; Omar, Khairuddin.

In: Applied Engineering in Agriculture, Vol. 30, No. 2, 2014, p. 325-334.

Research output: Contribution to journalArticle

@article{e248ee5208514fedae07dec9eb251e4f,
title = "Paddy abnormality vision recognition tool based on multi-layered mamdani fuzzy modeling",
abstract = "Conflicts often arise between farmers and experts regarding the solution for handling issues on paddy abnormality. Although multiple and serial courses are periodically conducted by the experts, yet the farmers have different ideas when dealing with paddy diseases. The disagreement regarding paddy diseases has made it difficult for Malaysia to meet its goal of producing 20 metric tons/ha. Hence, the objective of this research is to propose an automatic paddy abnormality visual recognition tool. A multi-layered Mamdani fuzzy modeling approach has been used in classifying three common paddy crop diseases, namely Blast, Brown Spot, and Narrow Brown Spot. Healthy leaf class is also included as a controlled parameter. The fuzzy classifier is designed to consist of four fuzzy membership functions by incorporating the characteristics of the lesion from the image gained from the texture analysis. Blob analysis is used in extracting features for the recognition of the lesion shape. Color charts are constructed with the help of an expert based on the RGB color space for the leaf color, lesion spot color, and lesion boundary color. In assessing a lesion, the best color match to the color charts is determined by using the Euclidean distance. An experiment is conducted where the performance of the proposed approach is compared to other classifiers which are ID3, Multi-layer Perceptron, Na{\"i}ve-Bayes and Decision Table/Na{\"i}ve-Bayes. This article proves that the proposed approach has the highest accuracy rate when compared to the other classifiers which is 85.80{\%}, with 87.01{\%} sensitivity and 81.24{\%} specificity rate, respectively.",
keywords = "Crop disease, Inspection, Multi-layered mamdani fuzzy modeling, Pattern recognition, Texture analysis",
author = "{Sheikh Abdullah}, {Siti Norul Huda} and Yusoff, {Raihan Afifah Mohamed} and Zainal, {Noor Faridatul Ainun} and Saad Abdullah and {Abu Bakar}, Azuraliza and Khairuddin Omar",
year = "2014",
doi = "10.13031/aea.30.10077",
language = "English",
volume = "30",
pages = "325--334",
journal = "Applied Engineering in Agriculture",
issn = "0883-8542",
publisher = "American Society of Agricultural and Biological Engineers",
number = "2",

}

TY - JOUR

T1 - Paddy abnormality vision recognition tool based on multi-layered mamdani fuzzy modeling

AU - Sheikh Abdullah, Siti Norul Huda

AU - Yusoff, Raihan Afifah Mohamed

AU - Zainal, Noor Faridatul Ainun

AU - Abdullah, Saad

AU - Abu Bakar, Azuraliza

AU - Omar, Khairuddin

PY - 2014

Y1 - 2014

N2 - Conflicts often arise between farmers and experts regarding the solution for handling issues on paddy abnormality. Although multiple and serial courses are periodically conducted by the experts, yet the farmers have different ideas when dealing with paddy diseases. The disagreement regarding paddy diseases has made it difficult for Malaysia to meet its goal of producing 20 metric tons/ha. Hence, the objective of this research is to propose an automatic paddy abnormality visual recognition tool. A multi-layered Mamdani fuzzy modeling approach has been used in classifying three common paddy crop diseases, namely Blast, Brown Spot, and Narrow Brown Spot. Healthy leaf class is also included as a controlled parameter. The fuzzy classifier is designed to consist of four fuzzy membership functions by incorporating the characteristics of the lesion from the image gained from the texture analysis. Blob analysis is used in extracting features for the recognition of the lesion shape. Color charts are constructed with the help of an expert based on the RGB color space for the leaf color, lesion spot color, and lesion boundary color. In assessing a lesion, the best color match to the color charts is determined by using the Euclidean distance. An experiment is conducted where the performance of the proposed approach is compared to other classifiers which are ID3, Multi-layer Perceptron, Naïve-Bayes and Decision Table/Naïve-Bayes. This article proves that the proposed approach has the highest accuracy rate when compared to the other classifiers which is 85.80%, with 87.01% sensitivity and 81.24% specificity rate, respectively.

AB - Conflicts often arise between farmers and experts regarding the solution for handling issues on paddy abnormality. Although multiple and serial courses are periodically conducted by the experts, yet the farmers have different ideas when dealing with paddy diseases. The disagreement regarding paddy diseases has made it difficult for Malaysia to meet its goal of producing 20 metric tons/ha. Hence, the objective of this research is to propose an automatic paddy abnormality visual recognition tool. A multi-layered Mamdani fuzzy modeling approach has been used in classifying three common paddy crop diseases, namely Blast, Brown Spot, and Narrow Brown Spot. Healthy leaf class is also included as a controlled parameter. The fuzzy classifier is designed to consist of four fuzzy membership functions by incorporating the characteristics of the lesion from the image gained from the texture analysis. Blob analysis is used in extracting features for the recognition of the lesion shape. Color charts are constructed with the help of an expert based on the RGB color space for the leaf color, lesion spot color, and lesion boundary color. In assessing a lesion, the best color match to the color charts is determined by using the Euclidean distance. An experiment is conducted where the performance of the proposed approach is compared to other classifiers which are ID3, Multi-layer Perceptron, Naïve-Bayes and Decision Table/Naïve-Bayes. This article proves that the proposed approach has the highest accuracy rate when compared to the other classifiers which is 85.80%, with 87.01% sensitivity and 81.24% specificity rate, respectively.

KW - Crop disease

KW - Inspection

KW - Multi-layered mamdani fuzzy modeling

KW - Pattern recognition

KW - Texture analysis

UR - http://www.scopus.com/inward/record.url?scp=84902961647&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84902961647&partnerID=8YFLogxK

U2 - 10.13031/aea.30.10077

DO - 10.13031/aea.30.10077

M3 - Article

AN - SCOPUS:84902961647

VL - 30

SP - 325

EP - 334

JO - Applied Engineering in Agriculture

JF - Applied Engineering in Agriculture

SN - 0883-8542

IS - 2

ER -