Introduction

Abstract

Remote sensing has become an indispensable tool for the earth observation, which provides the data in a pictorial form. Hyperspectral remote sensing images capture the reflectance response of the scene in the form of a few hundred spectral bands. Hyperspectral data have proved to be a highly useful resource due to their ability of providing a robust, accurate, and detailed information. A visualization of this huge volume of data over a standard tri-stimulus display is, however, a challenging problem as the hyperspectral data contain far more bands than those can be displayed on a standard display system. Therefore, one has to go through the entire set of bands to visualize the contents of the data. Additionally, the observer needs to co-register disparate information across the bands. This process is time consuming, inconsistent, and unreliable. Fusion provides an effective solution to the problem of visualization of hyperspectral image, where one obtains a single image representing most of the features of the input through an appropriate combination of the features across various hyperspectral bands. It also eliminates the need for going through the entire sequence of bands as the single fused image captures most of the important features. With some awareness of the scene contents, a human analyst may initiate data-specific processing algorithms, leading to significant computational savings. However, the image fusion problem is quite challenging as a huge amount of data from a few hundreds of bands have to be efficiently merged. This monograph discusses four techniques for visualization-oriented fusion of hyperspectral images using different methodologies. Apart from these techniques, this monograph also discusses band selection schemes in detail which may be employed to speed-up the fusion process. A dedicated chapter for evaluation of various fusion methodologies has also been provided. The current chapter provides an overview of hyperspectral imaging, and discusses the importance of visualization-oriented fusion of hyperspectral images.