基于流形学习算法的玉米叶部病斑图像降维与聚类分析

Analysis of clustering and dimensionality reduction of the corn leaf disease image based on manifold learning algorithm

流形学习算法是当今广泛使用的降维算法,但在玉米病害识别方面应用较少。本文通过利用Isomap、LLE和LE三种流形学习算法对玉米病斑图像进行降维对比研究,并运用K-means和K-medoids两种聚类算法对降维后的数据进行聚类分析,其中LLE算法在玉米叶部病斑图像的降维效果更好,能够为病害识别提供保障。

Manifold learning algorithm is widely used in dimensionality reduction, but it has fewer applications in the recognition of maize disease. Through using three calculating methods, Isomap, LLE and LE, the comparative study about the dimensionality reduction for maize disease images was made, and the data after dimension reduction was analyzed by using two clustering algorithms K-means and K-medoids. The LLE algorithm is much better in the dimension reduction effect and can provide guarantees for disease identification.