Enhanced pixel purity index algorithm to find the index position of the pure pixels in hyperspectral images

Abstract

This paper proposes an enhanced pure pixel index (PPI) algorithm for hyperspectral imaging. The conventional PPI algorithm uses random skewers in the process of finding the pure pixel indexes. The randomness in generating skewers leads to iterate the process multiple number of times to pick the most repeating pixel. The iteration involved in the process increases the computational time of the algorithm. The research contribution in this paper is that, the randomness in generating the skewers has been eliminated in the enhanced PPI algorithm by generating skewer by correlating the skewer with the given hyperspectral image. The proposed algorithm has reduced the computational complexity as well.