Methods of Spectral Matching of Heterogenous Remote Sensing Images
摘要
Abstract
Several methods for spectral matching of remote sensing data, including pixel-by-pixel linear method, pixel-by-pixel nonlinear method, and generalized nonlinear method, are proposed. The pixel-by-pixel nonlinear and generalized nonlinear methods are implemented using neural networks. Methods are compared using Landsat-8 and Sentinel-2 images. According to experimental results, the generalized nonlinear method for spectral matching of remote sensing images demonstrates the best matching quality.