An End-to-End Hyperspectral Image Classification Method Using Deep Convolutional Neural Network With Spatial Constraint

Hyperspectral image (HSI) classification is of vital importance in remote sensing-related applications. Various approaches, including the recently popular convolutional neural network (CNN)-based models, are proposed to tackle the problem of exploitation of the spatial and spectral features in the HSIs for the use of training classifier. In this letter, we design a simple but innovative end-to-end deep U-net-based model for HSI classification task. Unlike the previous CNN based models that mainly use CNN for spatial feature extraction and process the HSI data locally in small patches, our model takes the whole HSI as network input directly and outputs the predicted classes corresponding to each pixel location. Classification loss in the train data set and spatial constraint loss for the predicted result are combined as the loss function in the training stage to learn the mapping from HSI data to classification map and enhance the spatial continuity and consistency of the predicted result. Benchmark HSI data sets are used to evaluate the performance of the proposed method. Experimental results show that our model can achieve promising results comparing with the existing CNN-based methods.

I. Introduction

Hyperspectral image (HSI) classification plays an important role in many remote sensing-related applications, such as precision agriculture, mineralogy, and so on. HSI is a 3-D volume data with 2-D for geometrical or spatial structure, and the other 1-D for channel information. HSI contains information about the object class in each location on the land based on different properties of electromagnetic radiation at different wavelengths [1]. The task of HSI classification aims at classifying each pixel with a 1-D spectrum into the correct class via training on a limited manually labeled samples by exploiting the spatial and spectral information of the HSI data.