A fuzzy inference system for diagnosing oil palm nutritional deficiency symptoms

Muhammad H. Asraf, Nur K.A. Dalila, Amar Z.A. Faiz, Siti N. Aminah, M. T. Nooritawati

Research output: Contribution to journalArticlepeer-review

6 Citations (Scopus)


Automated monitoring of nutrient deficiencies by computers provides more accurate and precise information on current plant health status, which improves the effectiveness of fertilization management on large scale agricultural projects. In comparison to the traditional method which involves judging the deficiencies by personal observation, it is naturally more costly to hire agricultural experts who are more experienced. Besides, judgments by observations are also prone to misinterpretations, which will lead to incorrect applications of fertilizers. In this paper, the automatic detection via Fuzzy Inference System (FIS) is introduced as an automatic classification system for identifying the type of nutrient deficiencies in plant by emulating the judgment made by an agricultural expert when observing the state of the plant leaflets. The main objective of this study is to propose a design of Mamdani-FIS that models the classification of the deficiency type and severity. Oil palm tree (scientific name-Elaeis Guineensis Jacq.) were used as the study case for this paper, and leaflet samples are collected which consists of healthy leaflet and nutrient deficient leaflet that have been preprocessed to extract the inference via the fuzzy logic system. This involves the development of the fuzzy rule base, fuzzification and defuzzification process from the extracted features retrieved from leaflet images. The inputs of the membership functions of FIS consist of the number of red pixels, entropy and correlations. Three types of the nutrient deficiencies which are nitrogen, potassium and magnesium, and healthy leaflet condition was selected as the output of membership functions. The implementation of the system demonstrates a promising outcome, with classification accuracy confirmed to 82.67%. The performance of FIS is further evaluated via sensitivity, positive predictive value (ppv) and negative predictive value (npv) calculation which shows desirable rate of above 85%.

Original languageEnglish
Pages (from-to)3244-3250
Number of pages7
JournalARPN Journal of Engineering and Applied Sciences
Issue number10
Publication statusPublished - 1 May 2017


  • Classification
  • Fuzzy inference system
  • Fuzzy logic
  • Leaf recognition
  • Leaflet
  • Nutrient deficiencies
  • Oil palm


Dive into the research topics of 'A fuzzy inference system for diagnosing oil palm nutritional deficiency symptoms'. Together they form a unique fingerprint.

Cite this