DEVELOPMENT OF A SOFTWARE FOR REDUCTION OF HYPERSPECTRAL REMOTE SENSING DATA

Abstract

1-lyperspectral imagery is a new class of remote sensing data that can be used for identifying different surface materials. Airborne or satellite imaging spectrometers record reflected energy from the surface material into very narrow contiguous wavelength region. A hyperspectral data carries very detailed spectral information in hundreds of contiguous narrow spectral bands. This substantial dimensionality requires techniques which differ from traditional imagery analysis. One such approach is offered by reducing the dimensionality of hyperspectral data using Fractal based algorithm. The focus of the study is to reduce the dimensionality of hyperspectral data using different Fractal based algorithms. Fractal based dimensionality reduction algorithm reduces the overall dimension of a hyperspectral data by carrying out the reduction in SRC of all the pixels and sampling it in a fewer number of points. Four fractal based dimensionality reduction techniques are applied over an available hyperspectral image of Airborne Visible/Infrared Imaging Spectrometer (AVIRIS). Strengths and weaknesses of each method are developed. In general. the Multi-resolution box counting method and Line spacing method gave good results but suffers from high computational complexity. Sevcik's method is computationally efficient but requires large number of data sets of SRC.