Plant leaf classification based on weighted locally linear embedding

Abstract

Locally linear embedding (LLE) is effective in discovering the geometrical structure of the data. But when it is applied to real-world data, it shows some weak points, such as being quite sensitive to noise points and outliers, and being unsupervised in nature. In this paper, we propose a weighted LLE. The experiments on synthetic data and real plant leaf data demonstrate that the proposed algorithm can efficiently maintain an accurate low-dimensional representation of the noisy manifold data with less distortion, and acquire higher average recognition rates of plant leaf compared to other dimensional reduction methods.