Classifier Based on Cloud Model and Its Application
Lijuan Shi, Wen Youxian, Xingang Xie · 2009
The cloud model is good at bridging the gap between qualitatives and quantities, and so it was applied to evaluating mildew degree of rice seeds based on machine vision. A symmetric cloud and two asymmetric clouds were designed to express three qualitative concepts and the relationship between them. These three qualitative concepts represent different mildew degree including non-mildew, spot mildew and severe mildew. The mathematical property of each qualitative concept was described by a group of digital characteristics. After color features which can reflect the changes of diseased rice seeds were extracted from images, a cloud classifier was developed to classify the mildewed seeds elastically on the basis of cloud generators which implemented mapping between qualitativeness and quantities. Compared with the current rigid classifying methods, the cloud classifier was in conformity with the real distribution of data and simulated human thinking in qualitative way. An experiment was conducted to test on the classifier based on cloud and another classifier based on neural network. The results showed that the classification accuracy of cloud classifier was higher than the classification accuracy of neural network.