Comparative Analysis Based On Machine Learning And Deep Learning For Hyperspectral Image Classification
Hyperspectral image (HSI) contains continuous spectral bands acquired by hyperspectral sensors. Due to the high dimensional of spectral bands, distinguish the materials and objects is not easy in HSI. This work analysis the performance of the hyperspectral image classification based on machine learning and deep learning methods. There are five classification methods used in this paper. In Machine Learning, K-Nearest Neighbors (KNN), Navie Bayes and Back Propagation Network (BPN) are utilized. In Deep Learning, Convolutional Neural Network (CNN) and Bi-Long short Term Memory (Bi-LSTM) Network are used. These methods are used for HSI classification. A comparative study is discussed and evaluated on two bench mark dataset such as Pavia University and Indian Pines dataset. The experimental result provides the good classification accuracy for CNN and Bi-LSTM.