Due to the similar colors,it is difficult to distinguish between green peppers and pepper leaves in ordinary RGB im-ages.In order to provide technical support for picking green peppers in the field,it is urgent to explore the identification method of green peppers in the field.In this paper,a hyperspectral imager is used to scan field green peppers to obtain hyperspectral data.Af-ter the original data is calibrated by lens and reflectance,the data is normalized,and then the principal component analysis method is used to extract the spectral data characteristics of peppers,principal components.After the analysis,the four principal compo-nents with the largest contribution rate are obtained,and they are divided into four groups,each with three components,and each group of peppers is identified by the support vector machine and the BP neural network model.The results show that compared with SVM,PCA123-BPNN can significantly improve the recognition rate of pepper,the accuracy rate is 92.14%,and the recognition ef-fect is better.The PCA123-BPNN method proposed in this paper is intended to provide a reference for identifying and picking green peppers.
维普
万方数据