Hyperspectral Image Classification Based on Convolutional Neural Network

The deep learning is of great interest recently. MIT Technology Review ranked deep learning as the top 10 technology breakthrough in 2013. The multi-layered structure of Deep Learning Network enables it to learn deeper features of data and improve the accuracy of data identification and classification. Hyperspectral image classification technology is the most important research direction in hyperspectral remote sensing, and has been widely used in military and civil fields. It has become a trend to apply deep learning to hyperspectral image classification. But deep learning is data-driven and requires a lot of data, however, the hyperspectral data we deal with is a small data set. In the classification of small data sets, compared to the traditional classification methods, classification of the deep learning is more effective. In this paper, we study the convolution neural network which is excellent in the field of image processing, and compare the effect of hyperspectral image classification with the traditional Classification method.