Performance analysis of clustering method based on crop pest spectrum

In China, the crop diseases and insect pests are the main causes of output reduction and quality decline of crops. Through inspection of crop insects, we can prevent the pests in a timely and effective manner. The visible-near infrared (VNIR) spectral reflectance can intuitively reflect the growth, disease and insect pests information of crops, and through analysis of the crop's reflectance spectrum, we can detect and identify the crop pests. Clustering analysis is an important multivariable statistic and analysis method, and with the unsupervised learning method, we can effectively detect and classify the spectra of crop pests. In this paper, by using the spectral acquisition device designed by us, we collected three types of pests spectra on fresh broad bean leaves in a laboratory environment. We propose a scheme to analyze the clustering performance of crop pests spectra with the K-Means and the FCM clustering methods, and Matlab 2012b was used to realize the two clustering algorithms and analyze these clustering results. The experiment results show that the FCM clustering method has a better rate of identification, while the K-means clustering method has higher execution efficiency.