Hyperspectral unmixing based analysis of forested areas

This study aims to extract the planted regions in partially forested area by analyzing the hyperspectral remote sensing images acquired with airborne platforms. The proposed study utilizes the endmember signatures obtained from hyperspectral unmixing algorithms in order to classify the image pixels. The classification algorithm selects the endmember with highest spectral vegetation characteristic, and associates this endmember with the planted area pixels. The algorithm is tested on a scene covering METU Ankara campus area that is acquired by high resolution hyperspectral push-broom sensor operating in visible and NIR range of the electromagnetic spectrum on October, 22 2014.