Abstract
Satellite and airborne sensor images are useful for the monitoring of the vegetation states, such as crown cover state and crop growth. However, image data obtained through an optical sensor situated at high-altitude inevitably include mixels. Therefore, unmixing method, which estimates both the pure spectra and the coverage of endmembers simultaneously, is required in order to distinguish the qualitative spectral changes due to the chlorophyll quantity or crop variety, from the quantitative coverage change.
We have proposed an ICA (Independent Component Analysis) aided unmixing for periodically distributed hyperspectral data in agricultural land, and demonstrated with simulated mixel data that the technique enables us to estimate the pure spectra and coverage of crop and soil simultaneously even when the mixed spectra in agricultural land include vegetation covering fluctuation, and additive noise such as thermal sensor noise and atmospheric noise.
In this paper, we apply our ICA aided unmixing of agricultural land to a Japanese persimmon orchard in airborne hyperspectral sensor image and show the ability of the method against the actual remote sensing data.
