Color grading of cotton-grading (part II)

摘要

In this part of the series, two color grading system were developed using expert system and neural networks. Both grading systems have two modes of operation, classification mode and training mode. In the training mode, the expert system can be trained by a statistical method based on Bayes' theorem or genetic algorithm. For neural network approach, the grading system can be trained by backpropagation algorithm or probabilistic neural network. Using 100 cotton samples from USDA, the agreement with classer can be improved from the original 50percent from HVI grading to 86percent-100percent depending on the training method and the training samples. The relative contributions of each measurement on color grading were also investigated using stepwise discriminate analysis.