Functional Data Analysis: An Initiative Approach for Hyperspectral Data

Abstract

This paper summarizes initiative approache to analyze spectral variation and distribution for given spectral range of hyperspectral data using Functional Data Analysis (FDA.) The FDA can be used to analyse hyperspectral data which restrain information as curves or functions. Vertical shift and horizontal shift are two examples of directions of variation, or modes of variation in hyperspectral data. Shifts of curves in certain wavelength obtained from remote sensing are often used as a key indicator to obtain information from a target. As hyperspectral data use radiance or reflectance defined as the ratio of intensity of light received or reflected from the target to the intensity of the light incident on it, the radiance or reflectance values are then used to derive information about the composition of samples. In this paper, FDA method is introduced to present an alternative analysis in hyperspectral data as a function of spectral reflectance. distribution and variance of hyperspectral radiance or reflectance data.