Comparison of Several Hyperspectral Image Fusion Methods for Visualization

Abstract

Hyperspectral image visualization is an important research aspect in hyperspectral image fusion. This paper compared four typically used hyperspectral image visualization methods: method based on bilateral filter, method based on Principal Component Analysis (PCA), method based on independent component analysis (ICA) and method based on optimization. Fusion framework and scheme are explained briefly. Two sets of images obtained by AVIRIS and ROSIS sensors are used in our experiments, and four statistical assessment parameters, namely variance, entropy, average gradient and fusion factor are adopted to comparatively analyze the fusion results. The comparison results show that the effects of bilateral filter method, PCA method and optimization method are similar, and they are superior to ICA method.