Unsupervised band selection based on artificial bee colony algorithm for hyperspectral image classification

Abstract Hyperspectral image (HSI), with hundreds of narrow and adjacent spectral bands, supplies plentiful information to distinguish various land-cover types. However, these spectral bands ordinarily contain a lot of redundant information, leading to the Hughes phenomenon and an increase in computing time. As a popular dimensionality reduction technology, band/feature selection is indispensable for HSI classification. Based on improved subspace decomposition (ISD) and the artificial bee colony (ABC) algorithm, this paper proposes a band selection technique known as ISD–ABC to address the problem of dimensionality reduction in HSI classification. Subspace decomposition is achieved by calculating the correlation coefficients between adjacent bands and using the visualization result of the HSI spectral curve. The artificial bee colony algorithm is first applied to optimize the combination of selected bands with the guidance of ISD and maximum entropy (ME). Using the selected band subset, support vector machine (SVM) with five-fold cross validation is applied for HSI classification. To evaluate the effectiveness of the proposed method, experiments are conducted on two AVIRIS datasets (Indian Pines and Salinas) and a ROSIS dataset (Pavia University). Three indices, namely, overall accuracy (OA), average accuracy (AA) and kappa coefficient (KC), are used to assess the classification results. The experimental results successfully demonstrate that the proposed method provides good classification accuracy compared with six other state-of-the-art band selection techniques.