CLASSIFICATION OF PADDY WEED LEAF USING NEURO-FUZZY METHODS

Abstract

Paddy weed appears to be one of the many visible threats to paddy crop production and subsequently farmers’ income. It is for this reason that the growth of paddy weeds in paddy fields should be controlled as it results in a significant decrease of paddy yields. However, farmers might have limited knowledge on weed types, and are thus unable to identify and determine the right prevention methods. This paper presents classification methods for paddy weeds through the leaf shape extraction and applies neuro-fuzzy methods for recognizing the types of weeds. The types being focussed are the Sphenoclea zeylanica, Ludwigia hyssopifolia and Echinochloa crus-galli. The developed e-prototype methods would be able to classify paddy weeds with 83.78% accuracy. Hopefully, the findings in this study would assist farmers and researchers in increasing their paddy yields and eliminating weed growth respectively. The production of paddy in Malaysia would eventually be improved with the proposed methods, which can be considered as a technology advancement in the field of paddy production.