Min-max representation of features for grading cured tobacco leaves

In this paper, a new model to grade cured tobacco leaves using symbolic data is proposed. Color features such as average and standard deviation of hue of tobacco leaves are extracted using the Munsell color system. The extracted features of each grade are aggregated and represented in the form of inter-valued type data and stored in the knowledgebase. For a given test sample of a cured tobacco leaf, the features which are of crisp type are extracted and compared with the representatives stored in the knowledgebase. A symbolic classifier has been used for comparison purpose. In order to corroborate the efficacy of the proposed model, experiment has been conducted on real dataset consisting of 887 sample images of 12 different grades of cured tobacco leaves. The effect of grading accuracy under varying size of database has been studied and achieved an average grading accuracy of about 91%. A qualitative comparative analysis of the proposed model with other state of the art models is presented to study the elegance of the proposed model.