Investigation Of Hyperspectral Imaginary Data For Mineral Detection

Mohd Yusof, Umi Farrahhanim (2018) Investigation Of Hyperspectral Imaginary Data For Mineral Detection. Project Report. UTeM, Melaka, Malaysia. (Submitted)

Text (24 Pages) Investigation Of Hyperspectral Imaginary Data For Mineral Detection.pdf - Submitted Version Download (435kB)

Abstract

Hyperspectral unmixing (HU) is a very potential and increasingly popular preprocessing step for a wide range of hyperspectral application. Besides, it one of the important technique for hyperspectral data exploitation. Due to low spatial resolution of Hyperspectral cameras, microscopic material mixing, and multiple scattering, spectra measured by HSCs are mixtures of spectra of materials in a scene. Thus, accurate estimation is required using the hyperspectral unmixing algorithms. The linear mixing model is considered that assume hyperspectral image represented in the linear combination. However, this algorithm can perform all the chain involved in the hyperspectral unmixing process. The first chain is to the identification a set of pure spectral signature of the endmember. The second chain is to estimate the fractional abundances for each endmember for each pixel of the scene. These chains require an algorithm that highly desirable to avoid the propagation of error. The project is performed on a real HIS dataset as well on Airborne cuprite areas. Hence, this project attempt to propose a robust, stable, tractable, and accurate unmixing algorithm. The effectiveness of proposed method is confirmed through comparison with other algorithms, calculate the performance of unmixing by using spectral angle distance (SAD) under different noise level.

Actions (login required)

View Item