Affiliations: [a] College of Mechanical & Electronic Engineering, Northwest A&F University, Shaanxi, P.R.China | [b] Shaanxi Polytechnic Institute, Shaanxi, P.R.China | [c] School of Computer Science, University of Nottingham, Nottingham, NG81BB, UK
Correspondence:
[*]
Corresponding author. He Dongjian, College of Mechanical & Electronic Engineering, Northwest A&F University, Shaanxi, 712100 P.R.China. hdj168@nwsuaf.edu.cn
Abstract: Multiple features such as the margin, the shape and the texture of plant leaves are of great importance for classification of plant species, as they are often regarded as the unique features to identify plants. In this paper, we study the performance of a recently proposed semi-supervised fuzzy clustering algorithm with feature discrimination for leaf classification, based on features generated by principal component analysis of color images. The method outlines a basic framework for judging the weights of different features by adopting multiple feature matrixes obtained from the initial images as input data and the clustering results of the proposed clustering algorithm as output data to distinguish dissimilarities between various leaves. Real leaf images are employed to evaluate its performance and the experiment demonstrates that these results suggest that the margin feature, the shape feature and combination feature especially the margin feature and combination feature may be the best choice for leaf classification.
