Improving N-Finder technique for extracting endmembers

Abstract

N-FINDER algorithm is widely used for endmember extraction. One of the disadvantages of N-FINDER is that its implementations take long run time due to the relatively large computational complexity of N-FINDER. Successfully reducing the size of the input data set -the hyperspectral image - that the algorithm works on can reduce the overall run time of the algorithm. A method for successfully selecting the proper sample of the data set to work on is provided in this paper. Using this reduction technique, a faster and statistically more accurate version of N-FINDER is presented.