A novel fuzzy LBP based symbolic representation technique for classification of medicinal plants

In this paper, a novel fuzzy LBP model for extracting texture features from medicinal plant leaves is proposed. The proposed method is invariant to image transformations and independent of any threshold. Concept of hierarchical clustering based on inconsistency coefficient is used to produce natural clusters for a particular species capturing intra-class variations due to environmental conditions and acquisition system. Interval valued type symbolic feature vector is used to represent each cluster effectively. Thus the proposed system suggests choosing multiple representatives for each species to make the representation more effective and robust. A chi-square distance measure is used to establish matching between the test and reference feature vectors of plant leaves and a nearest neighbor classification technique is used to classify an unknown test sample of medicinal plant leaf. Extensive experiments are conducted to demonstrate the efficacy of the proposed model on our own data set and other publically available leaf datasets. Results of the proposed work has been compared with the contemporary work and found to be superior.