Multi-spectral Data Simulation Method Based on Hyperspectral Data

Abstract

Deep learning algorithms have been widely used in remote sensing image classification and recognition. Due to the difference in parameters such as spectral response functions of different sensors, a single sensor sample dataset cannot meet the needs of other sensors. Constructing sample datasets for different sensors will seriously increase the workload. To solve this problem, this paper proposes a multi-spectral data simulation method based on hyper spectral data and analyzes the accuracy of simulated data from typical surface spectrum curves, gray histograms and pixel errors. The results show that the simulation data is in good agreement with the original data, and the overall accuracy reaches 94.6%, which can solve the problem of deep learning sample dataset requirements.

Keywords: Data simulation, hyperspectral data, sample dataset

Cite this Article

X. Yang. Multi-spectral Data Simulation Method Based on Hyperspectral Data. Journal of Remote Sensing & GIS. 2019; 10(2): 43–51p.