Research of Plant-Leaves Classification Algorithm Based on Supervised LLE

Abstract

A new supervised LLE method based on the fisher projection was proposed in this paper, and combined it with a new classification algorithm based on manifold learning to realize the recognition of the plant leaves. Firstly，the method utilizes the Fisher projection distance to replace the sample's geodesic distance, and a new supervised LLE algorithm is obtained .Then, a classification algorithm which uses the manifold reconstruction error to distinguish the sample classification directly is adopted. This algorithm can utilize the category information better，and improve recognition rate effectively. At the same time, it has the advantage of the easily parameter estimation. The experimental results based on the real-world plant leaf databases shows its average accuracy of recognition was up to 95.17%.

DOI: http://dx.doi.org/10.11591/telkomnika.v11i6.2680