Accuracy of Fuzzy Logic-based Contamination Grading System on Abaca Tissue Culture

Abstract

This study aimed to support the coalition for abaca rehabilitation program of the different cooperating agencies of Southern Leyte with the use of intelligent systems in the grading of abaca tissue culture contamination. As the Southern Leyte State University Tissue Culture Laboratory has since used manual grading of contaminated specimen, the use of image acquisition technology can lessen the human effort involved in this activity. Fuzzy logic was used in the grading of the contamination. There were five inputs for the inference engine, namely red, green, blue, whitish and brownish. There were 62 rules identified covering the different combinations, and the triangular-shaped membership function was used for all the membership types of the inputs. Based on the result of the testing, it had an accuracy rate of 82.00% for the binary contamination in 50 sample specimens, and an over-all accuracy of 80.33% in multiclass contamination grading. The system could not match the precision of the human expert, but the model is comparable to that of the expected actual result, while minimizing human intervention in grading the contamination of tissue cultured abaca.